A Layout Design Decision-Support Framework
and Concept Demonstrator for Rural Hospitals
using Mixed Methods
by
Ingé Christine Kruger
Thesis presented in fulfilment of the requirements for the degree of Master of Engineering
(Industrial Engineering) in the Faculty of Engineering at Stellenbosch University
Supervisor: Ms Louzanne Bam
March 2017
Stellenbosch University https://scholar.sun.ac.za
DECLARATION
By submitting this thesis electronically, I declare that the entirety of the work contained therein is my
own, original work, that I am the sole author thereof (save to the extent explicitly otherwise stated),
that reproduction and publication thereof by Stellenbosch University will not infringe any third party
rights and that I have not previously in its entirety or in part submitted it for obtaining any
qualification.
Date: March 2017
Copyright © 2017 University of Stellenbosch
All rights reserved
ii
Stellenbosch University https://scholar.sun.ac.za
ABSTRACT
Layout design is an ever-present problem that has a significant effect on the operations of an
organisation, especially in the context of healthcare which deals with the lives of patients. It is a
complex problem that has long-term consequences and oftentimes competing objectives. Literature
has focused almost exclusively on using either quantitative or qualitative layout design methods for
designing layouts. This study develops a generic framework using both quantitative and qualitative
layout design methods that will guide the user to design a near optimal layout for a rural hospital
while taking into consideration the relevant laws and standards as well as the health outcomes of the
surrounding rural community. Rural and urban lifestyles, health, and illnesses differ in many ways.
General hospital design methods are therefore not necessarily appropriate for hospitals in these areas.
There is thus a need for a framework to be tailored for a rural community.
Following a mixed methods methodology, a systematic literature review of quantitative and
qualitative layout design methods along with hospital design considerations were conducted in order
to determine the most adequate methods for designing a hospital layout at the block diagram level of
detail. Furthermore, the commonalities and differences between rural and urban hospitals were
investigated including laws and standards relevant to hospital layouts.
The qualitative layout design methods involved different layout procedures and Muther’s Systematic
Layout Planning Procedure was found to be most adequate. Furthermore, hospital design
considerations such as patient-centeredness, efficiency, flexibility and expandability, sustainability,
and therapeutic environment were identified and linked with the quantitative layout methods. It was
also found that rural communities have different needs to urban ones with regard to access to medical
care, prominent illnesses, and attitudes towards health. The healthcare personnel shortages are
particularly problematic for rural communities.
The quantitative layout design methods involved layout models, solution methods (exact methods,
metaheuristics, and hybrid metaheuristics), and layout software. Using criteria of objectives,
assumptions, inputs, outputs, and hospital design considerations, the Quadratic Set Covering Problem
was determined to be the most appropriate model for designing a rural hospital block diagram layout.
It was deemed possible to integrate the quantitative and qualitative methods by embedding the
qualitative data into this quantitative model. The rural hospital design framework was developed
using Excel VBA and RStudio.
The framework was validated via two routes. Firstly, semi-structured interviews were conducted with
experts in the field, i.e. expert analyses. Secondly a case study of the Semonkong Hospital Project was
employed wherein the framework was applied successfully. The framework was deemed valid
according to both the expert analyses and the case study.
iii
Stellenbosch University https://scholar.sun.ac.za
OPSOMMING
Die uitleg van ‘n gebou het ‘n belangrike impak op die bedrywighede van ‘n organisasie – veral in die
konteks van gesondheidsorg waar daar met pasiënte se lewens gewerk word. Dit is ‘n ingewikkelde
probleem wat oor langtermyneffekte beskik en dikwels teenstrydige doelwitte. Die literatuur vir uitleg
ontwerpsmetodes het meestal gefokus op óf kwantitatiewe óf kwalitatiewe uitleg ontwerpsmetodes.
Hierdie studie ontwikkel ‘n generiese raamwerk wat beide van hierdie metodes gebruik om ‘n
gebruiker te lei om die uitleg van ‘n plattelandse hospital te ontwerp wat die gepaste wette en
standaarde en die gesondheid van die omliggende gemeenskap in ag neem. Landelike- en stedelike
gemeenskappe verskil in terme van hul lewenstyl, gesondheid en tipe siektes. Algemene uitleg
ontwerpsmetodes is dus nie noodwendig geskik vir ‘n plattelandse hospitaal nie. Daar is dus ‘n
behoefte om ‘n raamwerk te ontwikkel wat spesifiek is vir die uitleg van ‘n plattelandse hospitaal.
Hierdie studie volg ‘n gemengde metode benadering en ‘n sistematiese literatuurstudie is gevolglik
afsonderlik gedoen op kwantitatiewe- en kwalitatiewe uitleg ontwerpsmetodes met die doel om die
mees geskikte ontwerpsmetodes vir ‘n hospitaal uitleg te bepaal. Die verskille en ooreenkomste tussen
landelike- en stedelike hospitale was ook ondersoek. Hierdie sluit in wette en standaarde wat van
toepassing is op hospitaal uitlegte.
Die kwalitatiewe uitleg ontwerpsmetodes het verskillende uitleg prosedures ondersoek en dit is
gevind dat Muther se Sistematiese Uitleg Prosedure die mees geskik is vir die probleem van hierdie
studie. Daar is gevind dat die hoof ontwerpsoorwegings vir die uitleg van ‘n hospitaal pasiëntgesentreerdheid, doeltreffendheid, aanpasbaarheid, volhoubaarheid en terapeutiese omgewing is.
Daar is gevind dat landelike- en stedelike gemeenskappe verskil in terme van hul toegang tot mediese
sorg, prominente siektes, en hul houdings teenoor gesondheid. Een van die grootste probleme in
landelike hospitale was hul tekort aan personeel.
Die kwantitatiewe uitleg ontwerpsmetodes sluit uitleg modelle, oplossingsmetodes (presiese metodes,
metaheuristieke en hibriede metaheuristieke) en uitleg sagteware in. ‘n Kriteria van doelwitte,
aannames, insette, uitsette en hospitaal ontwerpsoorwegings was gebruik om die mees geskikte uitleg
model te kies naamlik: die ‘Quadratic Set Covering Problem’. Dit is gevind dat die kwantitatiewe- en
kwalitatiewe uitleg ontwerpsmetodes deur middel van ‘embedding’ geïntegreer kan word. Die uitleg
ontwerp raamwerk vir plattelandse hospitale was ontwikkel met behulp van Excel VBA en RStudio.
Die raamwerk is bekragtig deur twee roetes. Eerstens, semi-gestruktureerde onderhoude was gevoer
met kundiges in die velde van gesondheidsorg, plattelandse gemeenskappe en uitleg ontwerp.
Tweedens, die raamwerk is toegepas op ‘n gevallestudie van die Semonkong Hospitaal Projek. Albei
roetes dui daarop dat die raamwerk geldig is.
iv
Stellenbosch University https://scholar.sun.ac.za
ACKNOWLEDGEMENTS
“Not to us, Lord, not to us but to your name be the glory, because of your love and faithfulness.”
Psalm 115:1 NIV
I would like to express my sincerest appreciation to the following people who played an important
role during the course of this study:

First and foremost my supervisor, Ms Louzanne Bam, for her constant guidance, wisdom and patience
throughout this study. I wish you every success and happiness.

The Department of Industrial Engineering, for assisting me throughout my studies.

The Semonkong Hospital Project, for their time, assistance and advice.

My family, for their love and support.

Michael, your encouragement, support, and company has made time spent working a treasured memory.
v
Stellenbosch University https://scholar.sun.ac.za
TABLE OF CONTENTS
DECLARATION .............................................................................................................................................. ii
ABSTRACT ................................................................................................................................................... iii
OPSOMMING ................................................................................................................................................ iv
ACKNOWLEDGEMENTS ............................................................................................................................. v
LIST OF SYMBOLS ..................................................................................................................................... xii
LIST OF ACRONYMS .................................................................................................................................. xv
LIST OF FIGURES ......................................................................................................................................xvi
LIST OF TABLES ..................................................................................................................................... xviii
LIST OF ALGORITHMS ............................................................................................................................xxi
CHAPTER 1 INTRODUCTION ................................................................................................................... 1
1.1
Introduction .................................................................................................................................................................. 2
1.2
Background ................................................................................................................................................................... 2
1.2.1
Facility layout design ............................................................................................................................................ 2
1.2.2
Facility layout design literature ....................................................................................................................... 5
1.3
Problem definition...................................................................................................................................................... 5
1.4
Aim and objectives ..................................................................................................................................................... 5
1.5
Research design ........................................................................................................................................................... 7
1.6
Research methodology ............................................................................................................................................. 7
1.7
Research scope and nature ..................................................................................................................................... 9
1.8
Document outline..................................................................................................................................................... 10
1.9
Conclusion................................................................................................................................................................... 11
CHAPTER 2 QUANTITATIVE LAYOUT DESIGN METHODS .......................................................... 13
2.1
Introduction ............................................................................................................................................................... 14
2.2
Operations Research............................................................................................................................................... 14
2.3
Facility Layout Problem ........................................................................................................................................ 15
2.3.1
Quantitative methods for modelling and solving the Facility Layout Problem......................... 16
2.3.2
Quantitative methods software packages ................................................................................................. 20
2.3.3
Application of the facility layout problem to hospital layouts ......................................................... 20
2.4
2.4.1
Layout models ........................................................................................................................................................... 20
Classification of models .................................................................................................................................... 21
vi
Stellenbosch University https://scholar.sun.ac.za
TABLE OF CONTENTS
2.4.2
Quadratic Assignment Problem .................................................................................................................... 21
2.4.3
Quadratic Set Covering Problem................................................................................................................... 25
2.4.4
Linear Integer Programming Problem ....................................................................................................... 27
2.4.5
Mixed Integer Programming Problem ........................................................................................................ 30
2.4.6
Graph Theoretic Problem ................................................................................................................................ 32
2.4.7
Linear Continuous Model................................................................................................................................. 34
2.4.8
Linear Mixed Integer Model............................................................................................................................ 37
2.5
Solution methods ..................................................................................................................................................... 38
2.5.1
Exact methods ...................................................................................................................................................... 39
2.5.2
Metaheuristics ...................................................................................................................................................... 40
2.5.3
Hybrid metaheuristics ...................................................................................................................................... 55
2.6
Layout software ........................................................................................................................................................ 56
2.7
Analysis of methods ................................................................................................................................................ 59
2.7.1
Analysis of layout models ................................................................................................................................ 59
2.7.2
Analysis of solution methods ......................................................................................................................... 62
2.8
Conclusion................................................................................................................................................................... 63
CHAPTER 3 QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN
CONSIDERATIONS .................................................................................................................................... 64
3.1
Introduction ............................................................................................................................................................... 65
3.2
Layout types ............................................................................................................................................................... 65
3.3
Layout Procedures................................................................................................................................................... 67
3.3.1
Apple’s Plant Layout Procedure .................................................................................................................... 68
3.3.2
Reed’s Layout Procedure ................................................................................................................................. 68
3.3.3
Muther’s Systematic Layout Procedure ..................................................................................................... 69
3.3.4
Conclusion: Layout procedures ..................................................................................................................... 70
3.4
Department level analysis .................................................................................................................................... 70
3.4.1
Administration ..................................................................................................................................................... 72
3.4.2
Obstetrics................................................................................................................................................................ 73
3.4.3
Operating Suite..................................................................................................................................................... 74
3.4.4
Paediatric Unit ...................................................................................................................................................... 75
vii
Stellenbosch University https://scholar.sun.ac.za
TABLE OF CONTENTS
3.4.5
Outpatient unit ..................................................................................................................................................... 75
3.4.6
Laundry ................................................................................................................................................................... 76
3.4.7
Kitchen ..................................................................................................................................................................... 76
3.4.8
Sterilisation and Disinfection Unit ............................................................................................................... 77
3.4.9
Pharmacy ................................................................................................................................................................ 77
3.4.10
Emergency and Casualty Unit ................................................................................................................... 78
3.4.11
Acute Psychiatric Facility ............................................................................................................................ 78
3.4.12
Chronic Care Unit ........................................................................................................................................... 79
3.4.13
Rehabilitation Unit......................................................................................................................................... 80
3.4.14
Mortuary ............................................................................................................................................................ 80
3.4.15
Laboratory......................................................................................................................................................... 81
3.4.16
Radiology ........................................................................................................................................................... 81
3.4.17
Neonatal Intensive care Unit ..................................................................................................................... 82
3.4.18
Intensive Care Unit ........................................................................................................................................ 82
3.4.19
High Care Unit.................................................................................................................................................. 83
3.4.20
Resulting departmental relationships and flows .............................................................................. 83
3.5
Hospital design considerations .......................................................................................................................... 85
3.5.1
Patient-centeredness ......................................................................................................................................... 85
3.5.2
Efficiency ................................................................................................................................................................ 86
3.5.3
Flexibility and expandability .......................................................................................................................... 89
3.5.4
Sustainability ........................................................................................................................................................ 90
3.5.5
Therapeutic environment ................................................................................................................................ 90
3.5.6
Linking design considerations to quantitative methods .................................................................... 92
3.6
Conclusion................................................................................................................................................................... 94
CHAPTER 4 RURAL VERSUS URBAN HOSPITALS ........................................................................... 95
4.1
Introduction ............................................................................................................................................................... 96
4.2
Commonalities among rural and urban hospitals ...................................................................................... 96
4.2.1
General building constraints .......................................................................................................................... 97
4.2.2
Hospital layout constraints ............................................................................................................................. 97
4.2.3
Adherence to standards.................................................................................................................................... 98
viii
Stellenbosch University https://scholar.sun.ac.za
TABLE OF CONTENTS
4.3
Rural-specific constraints ..................................................................................................................................... 99
4.3.1
Healthcare personnel constraints ................................................................................................................ 99
4.3.2
Rural community considerations ...............................................................................................................101
4.4
Floor layout implications ....................................................................................................................................105
4.5
Conclusion.................................................................................................................................................................106
CHAPTER 5 PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS .......... 107
5.1
Introduction .............................................................................................................................................................108
5.2
Integration of quantitative and qualitative methods ..............................................................................108
5.3
Theoretical framework definition...................................................................................................................109
5.4
Real world problem ..............................................................................................................................................109
5.5
Selection of layout model....................................................................................................................................110
5.5.1
Analysis of layout model objectives ..........................................................................................................110
5.5.2
Analysis of layout model assumptions .....................................................................................................112
5.5.3
Analysis of layout model inputs ..................................................................................................................114
5.5.4
Analysis of layout model outputs ...............................................................................................................115
5.5.5
Analysis of design considerations ..............................................................................................................116
5.5.6
Selection of layout model ...............................................................................................................................117
5.5.7
Adjustments to layout model .......................................................................................................................118
5.6
Selection of solution method.............................................................................................................................118
5.7
Development of layout design framework ..................................................................................................119
5.7.1
Assumptions ........................................................................................................................................................119
5.7.2
Selection of programs......................................................................................................................................119
5.7.3
Framework steps...............................................................................................................................................120
5.7.4
Examples...............................................................................................................................................................131
5.8
Conclusion.................................................................................................................................................................137
CHAPTER 6 VALIDATION AND VERIFICATION ............................................................................. 139
6.1
Introduction .............................................................................................................................................................140
6.2
Verification ...............................................................................................................................................................140
6.3
Various routes to validation ..............................................................................................................................140
6.4
Subject-matter expert analysis.........................................................................................................................141
ix
Stellenbosch University https://scholar.sun.ac.za
TABLE OF CONTENTS
6.4.1
The process for validation via interviews ...............................................................................................141
6.4.2
Interviewee profile ...........................................................................................................................................141
6.4.3
Interviewee summary .....................................................................................................................................142
6.4.4
Validation questions ........................................................................................................................................143
6.4.5
Summary of feedback from experts ..........................................................................................................143
6.5
Case study .................................................................................................................................................................145
6.5.1
The Semonkong Hospital Project ...............................................................................................................146
6.5.2
Application of the framework to the case study ..................................................................................148
6.5.3
Feedback from the Semonkong Hospital Project .................................................................................156
6.6
Conclusion.................................................................................................................................................................157
CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS ............................................................. 158
7.1
Introduction .............................................................................................................................................................159
7.2
Research summary ................................................................................................................................................159
7.3
Research contributions .......................................................................................................................................160
7.4
Recommendations for future research .........................................................................................................161
REFERENCES ............................................................................................................................................ 162
APPENDICES............................................................................................................................................. 188
APPENDIX A: HOSPITAL LAYOUT CONSTRAINTS ..................................................................................................189
APPENDIX B: NATIONAL CORE STANDARDS ...........................................................................................................203
APPENDIX C: VERIFICATION OF THE FRAMEWORK ............................................................................................208
C.1 Hospital layout design questions verification ...............................................................................................208
C.2 Research objectives verification .........................................................................................................................208
C.3 Framework outputs verification.........................................................................................................................210
APPENDIX D: DOCUMENT FOR INTERVIEWS WITH EXPERTS.........................................................................211
D.1 Validation presentation .........................................................................................................................................211
D.2 Validation document and questionnaire ........................................................................................................244
D.3 Expert analysis feedback .......................................................................................................................................270
APPENDIX E: PROGRAM CODE .......................................................................................................................................279
x
Stellenbosch University https://scholar.sun.ac.za
TABLE OF CONTENTS
APPENDIX F: UPDATED FRAMEWORK .......................................................................................................................284
xi
Stellenbosch University https://scholar.sun.ac.za
LIST OF SYMBOLS
Symbol
A
𝑏𝑖𝑗𝑘𝑙
Meaning
𝑛 x 𝑛 matrix
monetary term of locating department 𝑖 at location 𝑗
𝑐𝑗𝑙
cost of transporting one patient between location j and location 𝑙
𝑐𝑖𝑗
cost of transporting one patient between location 𝑗 and location 𝑙
C
Cost matrix with elements [𝑐𝑖𝑗 ]
𝑑𝑗𝑙
distance between the centroids of locations 𝑗 and 𝑙 if facility 𝑖 is assigned to location
𝑗 and facility 𝑘 is assigned to location 𝑙
D
distance matrix with elements [𝑑𝑘𝑙 ] where 𝑑𝑘𝑙 represents the distance between
location 𝑘 and location 𝑙
𝑒𝑖𝑗
the double summation of the monetary term of locating department 𝑖 at location 𝑗
𝐸
a set of edges disjoint from 𝑉
𝐹
the set of pairs of departments which must not be adjacent in any feasible solution,
𝑓𝑖𝑘
flow of patients between department 𝑖 and department 𝑘
𝑔𝑖
the distance between the extreme horizontal sides of department 𝑖
𝑔𝑗
the distance between the extreme horizontal sides of department 𝑗
𝐺
a weighted graph with 𝑉 as a nonempty set of departments, 𝐸 as a set of edges
disjoint from 𝑉
𝑙
ℎ𝑖𝑘
horizontal distance between facilities 𝑖 and k when facility 𝑖 is to the left of facility 𝑘;
0 otherwise
𝑟
ℎ𝑖𝑘
horizontal distance between facilities 𝑖 and k when facility 𝑖 is to the right of facility
𝑘; 0 otherwise
ℎ𝑖
the horizontal distance between department 𝑖 and the v-axis
ℎ𝑗
the horizontal distance between department 𝑗 and the v-axis
𝐼(𝑖)
Set of candidate locations for department 𝑖
𝐽𝑖 (𝑗)
Set of blocks occupied by facility 𝑖 if it is assigned to location 𝑗
𝑘𝑖𝑗
minimum distance by which departments 𝑖 and 𝑗 are to be separated
𝑙𝑖
the distance between the extreme vertical sides of department 𝑖
𝑙𝑗
the distance between the extreme vertical sides of department 𝑗
𝑀
an arbitrarily large positive number
m
Number of locations of departments
𝑛
number of departments
𝑁
the set of pairs of departments which must be adjacent in any feasible solution,
𝑜𝑖𝑗
variable indicating if department 𝑖 is adjacent to department j. 1 if department 𝑖 is
xii
Stellenbosch University https://scholar.sun.ac.za
LIST OF SYMBOLS
adjacent to department 𝑗; 0 otherwise
𝑝𝑘𝑙
semi-net revenue from the operation of department 𝑘 at location 𝑖. In other words
the gross revenue less cost of primary inputs, but before subtracting transportation
cost of intermediate products between department
𝑞
number of blocks into which the total area occupied by all departments is divided
into
𝑟𝑖𝑗
the closeness rating indicating desirability of locating department 𝑖 adjacent to
department 𝑗
𝑣
𝑠𝑖𝑗
vertical clearance, i.e. the minimum distance by which departments 𝑖 and 𝑗 are to be
separated if they are positioned in opposite rows in the layout
ℎ
𝑠𝑖𝑗
the horizontal clearance, i.e. the minimum distance by which departments 𝑖 and 𝑗
are to be separated if they are positioned in the same rows in the layout
𝑡𝑖𝑗
the time required for a patient to move from department i to department j
𝑢𝑘𝑙
set of non-negative numbers that represent the required commodity flows in weight
units from department 𝑘 to department 𝑙. 𝑘 ≠ 1. l = 1, …, 𝑛
𝑉
𝑎
𝑣𝑖𝑘
nonempty set of departments
vertical distance between facilities 𝑖 and k when facility 𝑖 is above facility 𝑘; 0
otherwise
𝑏
𝑣𝑖𝑘
vertical distance between facilities 𝑖 and 𝑘 when facility 𝑖 is below facility 𝑘; 0
otherwise
𝑣𝑖
the vertical distance between department 𝑖 and h-axis
𝑣𝑗
the vertical distance between department 𝑗 and h-axis
𝑤𝑖𝑗
a function of the assignment of departments and the monetary term of locating
department 𝑖 at location 𝑗
X
𝑥𝑖𝑗
Permutation 𝑛 x 𝑛 matrix with elements [𝑥𝑖𝑗 ]
assignment of department 𝑖 to location j. 1 if department 𝑖 is at location 𝑗; 0
otherwise
𝑥𝑘𝑙,𝑖𝑗
1 if there is flow from location 𝑖 to location 𝑗 of the commodity which is supplied by
department 𝑘 to department 𝑙; 0 if there is no flow
𝑥𝑖𝑙
(𝑥̅𝑖 , 𝑦̅𝑙 )
𝑦𝑖𝑗𝑘𝑙
𝑧𝑖𝑗
1 if department 𝑖 is at location 𝑙; 0 otherwise
location of facility 𝑖
the integer variable of department 𝑖 at location 𝑗 in arrangement 𝑘 of location 𝑙
variable used to ensure that only one of the constraints hold so that the
departments do not overlap
𝛼𝑖𝑗
𝜃
the frequency of trips required between department 𝑖 and 𝑗
real-valued scaling parameter
xiii
Stellenbosch University https://scholar.sun.ac.za
LIST OF SYMBOLS
𝛾𝑖𝑗
fixed cost of locating and operating department 𝑖 at location 𝑗
xiv
Stellenbosch University https://scholar.sun.ac.za
LIST OF ACRONYMS
CA
Circulation Area
CF
Circulation Factor
CM
Circulation Multiplier
CS
Community Service
CSIR
Council for Scientific and Industrial Research
FLP
Facility Layout Problem
GA
Genetic Algorithm
GTP
Graph Theoretic Problem
IASC
Inter-Agency Standing Committee
IUSS
Infrastructure Unit Support Systems
LCM
Linear Continuous Model
LIPP
Linear Integer Programming Problem
LMIM
Linear Mixed Integer Model
LS
Local Search
MIPP
Mixed Integer Programming Problem
NA
Net Area
OECD
Organisation for Economic Co-operation and Development
QAP
Quadratic Assignment Problem
QSCP
Quadratic Set Covering Problem
UA
Usable Area
WHO
World Health Organisation
xv
Stellenbosch University https://scholar.sun.ac.za
LIST OF FIGURES
Figure 1: Hospital layout design steps and phases, adapted from Tompkins, White, Bozer and Tanchoco
(2010) ................................................................................................................................................................................................. 6
Figure 2: Research design mapping, adapted from Mouton (2011) ......................................................................... 7
Figure 3: Research methodology ............................................................................................................................................. 9
Figure 4: Outline of thesis chapters ..................................................................................................................................... 10
Figure 5: Modelling and solving methods for the Facility Layout Problem ........................................................ 16
Figure 6: Timeline of quantitative methods used to solve the Facility Layout Problem ............................... 19
Figure 7: Chosen layout models and solution methods for further analysis ...................................................... 19
Figure 8: Variables for Equal-sized Linear Continuous Model ................................................................................. 34
Figure 9: Variables for unequal-sized Linear Continuous Model ............................................................................ 34
Figure 10: Variables for unequal-sized Linear Mixed Integer Model .................................................................... 37
Figure 11: Behaviour of ants searching for an optimal path between the food and their nest adapted
from Dorigo, Maniezzo and Colorni (1996) ..................................................................................................................... 51
Figure 12: Particle Swarm Optimisation example adapted from Heppner and Grenander (1990).......... 54
Figure 13: Apple's Plant Layout Procedure adapted from Tompkins et al. (2010) ......................................... 68
Figure 14: Reed's Layout Procedure adapted from Tompkins et al. (2010) ...................................................... 69
Figure 15: Muther's Systematic Layout Procedure adapted from Tompkins et al. (2010) .......................... 69
Figure 16: South African hospital service package ....................................................................................................... 72
Figure 17: Relationship diagram of hospital departments ........................................................................................ 84
Figure 18: Flow diagram of hospital departments ........................................................................................................ 84
Figure 19 Rural versus urban comparison structure and objectives .................................................................... 96
Figure 20: Cause and effect diagram of rural-specific challenges faced by rural communities ...............101
Figure 21: Framework outline .............................................................................................................................................121
Figure 22: Example 1 output ................................................................................................................................................133
Figure 23: Example 2 output ................................................................................................................................................134
Figure 24: Example 3 output ................................................................................................................................................135
Figure 25: Example 4 output ................................................................................................................................................136
Figure 26: The interviewees' areas of expertise ..........................................................................................................142
Figure 27: The Semonkong Hospital Project's site space .........................................................................................148
Figure 28: Department relationships suggested and selected for the Jehovah Shammah Hospital .......150
Figure 29: Dimensions for the site of the Jehovah Shammah Hospital ...............................................................152
Figure 30: Possible locations of centres of the departments of the Jehovah Shammah Hospital ............153
Figure 31: Proposed layout of the Jehovah Shammah Hospital .............................................................................154
Figure 32: RStudio solution of the Jehovah Shammah Hospital using a larger site area.............................155
xvi
Stellenbosch University https://scholar.sun.ac.za
LIST OF FIGURES
Figure 33: Updated framework ...........................................................................................................................................284
xvii
Stellenbosch University https://scholar.sun.ac.za
LIST OF TABLES
Table 1: Quantitative Methods used to model and solve the Facility Layout Problem .................................. 16
Table 2 Layout models classified according to discrete and continuous ............................................................. 21
Table 3: General formulation of the Quadratic Assignment Problem (Heragu & Kusiak, 1987; Konak,
Kulturel-Konak, Norman & Smith, 2006) .......................................................................................................................... 22
Table 4: Koopmans and Beckmann formulation of the Quadratic Assignment Problem (Beckmann &
Koopmans, 1957; Heragu & Kusiak, 1987)....................................................................................................................... 23
Table 5: Trace formulation of the Quadratic Assignment Problem (Burkard et al., 2009; Commander,
2003) ................................................................................................................................................................................................ 25
Table 6: Quadratic Set Covering Problem (Heragu & Kusiak, 1987; Bester, 2006) ......................................... 26
Table 7: Linear Integer Programming Problem (Lawler, 1963; Heragu & Kusiak, 1987) ............................ 27
Table 8: Linear Integer Programming Problem (Love & Wong, 1976)................................................................. 29
Table 9: Mixed Integer Programming Problem (Heragu & Kusiak, 1987) .......................................................... 30
Table 10: Graph Theoretic Problem (Heragu & Kusiak, 1987) ................................................................................ 32
Table 11: Linear Continuous Model for equal-sized departments (Heragy & Kusiak, 1991)...................... 35
Table 12: Linear Continuous Model for unequal sized departments (Heragy & Kusiak, 1991) ................. 36
Table 13: Linear Mixed Integer Model (Heragy & Kusiak, 1991)............................................................................ 37
Table 14: Strengths and weaknesses of Simulated Annealing adapted from Dowsland and Thompson
(2012) and Vecchi (1983)........................................................................................................................................................ 45
Table 15: Strengths and weaknesses of Tabu Search adapted from Hanafi (2001) and Abido (2002) .. 47
Table 16: Strengths and weaknesses of Genetic Algorithms adapted from Rajeev and Krishnamoorthy
(1992) and Simpson, Dandy and Murphy (1994).......................................................................................................... 49
Table 17: Strengths and weaknesses of Ant Colony Optimisation adapted from Boswal, Poddar and
Shekhawat (2009)....................................................................................................................................................................... 52
Table 18: Strengths and weaknesses of Particle Swarm Optimisation adapted from Talbi (2009) and
Reynolds (1987) .......................................................................................................................................................................... 55
Table 19: Classification of computer-based algorithms adapted from Schiffauerova (2014) .................... 57
Table 20: Analysis of layout models .................................................................................................................................... 60
Table 21: Metaheuristics classification .............................................................................................................................. 62
Table 22: Strengths and weaknesses of a process layout adapted from Groover (2007), Karvonen et al.
(2007), and Ali (2015) .............................................................................................................................................................. 67
Table 23: Closeness rating scale for departmental relationships ........................................................................... 71
Table 24: Qualitative Hospital design considerations ................................................................................................. 85
Table 25: Non-critical and Critical Minimum Passing Spaces adapted from The Accident Compensation
Corporation (2011) .................................................................................................................................................................... 89
Table 26: Linking hospital design measures and layout models ............................................................................. 93
xviii
Stellenbosch University https://scholar.sun.ac.za
LIST OF TABLES
Table 27: Incorporation of hospital design methods in framework ...................................................................... 93
Table 28: Rural population size and the corresponding medical scheme coverage of provinces in South
Africa adapted from National Department of Health (2011) .................................................................................... 99
Table 29: Human resources and medical practitioners per province in South Africa adapted from
National Department of Health (2011) ............................................................................................................................100
Table 30: Unemployment rates and children living in poverty .............................................................................102
Table 31: Analysis of layout models' objectives ...........................................................................................................112
Table 32: Analysis of layout models' assumptions......................................................................................................113
Table 33: Analysis of layout model’s inputs ...................................................................................................................115
Table 34: Analysis of layout models' outputs ................................................................................................................116
Table 35: Analysis of layout models' design considerations ...................................................................................117
Table 36: Selection of the best layout model .................................................................................................................117
Table 37: Benchmarking hospital bed distribution adapted from Life Healthcare (2016)........................126
Table 38: Number of beds and approximate size of laundry adapted from Fourie, Sheared and Steyn
(2014) ............................................................................................................................................................................................128
Table 39: Number of beds and size of kitchen ..............................................................................................................128
Table 40: Number of beds and size of Sterilisation and Disinfection Unit adapted from Steyn and
Sheard (2004) .............................................................................................................................................................................129
Table 41: Example of Excel VBA output...........................................................................................................................130
Table 42: Example 1 relationships .....................................................................................................................................132
Table 43: Example 2 relationships .....................................................................................................................................133
Table 44: Example 3 relationships .....................................................................................................................................134
Table 45: Example 4 relationships .....................................................................................................................................136
Table 46: Validation questions ............................................................................................................................................143
Table 47 Interviewee feedback summary .......................................................................................................................144
Table 48: Suggested and selected number of beds for the Jehovah Shammah Hospital .............................149
Table 49: Summary of the sizes and costs of each department in the Jehovah Shammah Hospital .......150
Table 50: Dimensions and number of blocks required for the Jehovah Shammah Hospital .....................151
Table 51: Evaluation of the proposed layout for the Jehovah Shammah Hospital.........................................154
Table 52 Semonkong Hospital Project interviewees feedback summary..........................................................156
Table 53: Administration constraints ...............................................................................................................................189
Table 54: Obstetrics constraints .........................................................................................................................................189
Table 55: Operating Suite constraints ..............................................................................................................................191
Table 56: Paediatric Unit constraints ...............................................................................................................................191
Table 57: Outpatient department .......................................................................................................................................193
Table 58: Laundry constraints .............................................................................................................................................195
Table 59: Kitchen constraints ..............................................................................................................................................195
xix
Stellenbosch University https://scholar.sun.ac.za
LIST OF TABLES
Table 60: Sterilisation and Disinfection Unit constraints .........................................................................................195
Table 61: Pharmacy constraints .........................................................................................................................................196
Table 62: Emergency and Casualty Unit constraints ..................................................................................................196
Table 63: Acute Psychiatric Facility constraints ..........................................................................................................197
Table 64: Chronic Care Unit constraints ..........................................................................................................................198
Table 65: Rehabilitation Unit constraints .......................................................................................................................198
Table 66: Mortuary constraints...........................................................................................................................................200
Table 67: Laboratory constraints .......................................................................................................................................200
Table 68: Radiology constraints..........................................................................................................................................200
Table 69: Neonatal Intensive Care Unit constraints ...................................................................................................201
Table 70: Intensive Care Unit constraints .......................................................................................................................201
Table 71: High Care Unit constraints ................................................................................................................................202
Table 72: Facilities and infrastructure domain of the National Core Standards ............................................204
Table 73: Hospital layout design questions verification ...........................................................................................208
Table 74: Research objectives verification .....................................................................................................................208
Table 75: Framework outputs verification.....................................................................................................................210
xx
Stellenbosch University https://scholar.sun.ac.za
LIST OF ALGORITHMS
Algorithm 2.1: Iterative improvement local search algorithm adapted from Blum, Roli and Sampels
(2008) .............................................................................................................................................................................................. 43
Algorithm 2.2: Simulated Annealing algorithm outline adapted from Bertsekas (1999) ............................. 46
Algorithm 2.3: Tabu Search Algorithm outline adapted from Gendreau (2003).............................................. 47
Algorithm 2.4: Genetic Algorithm outline adapted from Rajeev and Krishnamoorthy (1992) .................. 49
Algorithm 2.5: Ant Colony Optimisation algorithm outline adapted from Dorigo and Stützle (2003) ... 52
Algorithm 2.6: Particle Swarm algorithm outline adapted from Rardin (1998) and Romero, Zamudio,
Baltazar, Mezura, Sotelo and Callaghan (2012).............................................................................................................. 55
xxi
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1
INTRODUCTION
“Measure twice, cut once.”
Davis (2010)
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
1
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
1.1 Introduction
To begin with, it is necessary to lay the foundations of the research and ensure its usefulness to the
greater body of knowledge. The purpose of this chapter is to introduce the problem of designing a
hospital layout for a rural community and therefore provide the rationale for this study. The primary
focus of this research is expressed in terms of the aim and objectives listed in Section 1.3. Thereafter,
the research design and methodology is discussed in order to achieve these research aims. To
conclude this introductory chapter, the outlines of each of the proceeding chapters are given.
1.2 Background
This section introduces facility layout design and the role it plays on the outcome of its organisation. A
hospital in particular poses a unique set of challenges and different design concerns. The focus is
placed on designing a hospital layout for a rural community. Lastly, facility layout literature is
introduced.
1.2.1 FACILITY LAYOUT DESIGN
There is a saying that one should “measure twice, cut once” (Davis, 2010). The planning phase of a
facility’s layout is considerably more important than the actual construction phase. For an
organisation to be effective and efficient, it is important to pay special attention to the facility layout.
Carr (2011) states that “a functional design can promote skill, economy, conveniences, and comforts; a
non-functional design can impede activities of all types, detract from quality of care, and raise costs to
intolerable levels.” An effective facility layout ensures a smooth and steady flow of production material,
equipment, and manpower at a minimum cost. It should consider the available space, final product or
service, safety of employees and facility as well as the convenience of operations (Management Study
Guide, 2015). The design of a facility, also known as the Facility Layout Problem (FLP), regards the
arrangement of units in a specific area that has to satisfy certain constraints and preferences while
optimising an objective function value. These preferences are oftentimes subjective, uncertain and
complex. According to Youssef, Sait and Ali (2003) the design of layouts are still vague and undefined.
Literature mainly focused on rigid and simplified frameworks that lack the appropriate methodologies
for using them (Ahmad, Basir, Hassanein & Azam, 2008).
Hospitals in particular are complex systems and their environment has a substantial impact on the
health, safety and overall well-being of patients as well as staff. However, safety and improving quality
is often not integrated into the design phase of hospital buildings (Barach & Dickerman, 2007).
According to Jacobs, Chase and Chase (2010) the layout of a hospital sets physical constraints on its
operations. The British Medical Association (2011) (BMA) found that “the architectural environment
can contribute to the treatment of patients and significantly affect their health outcomes.” More than
600 studies have shown that there is a direct link between healthcare design and medical
2
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
outcomes (Centre for Health Design, 2004). A few examples of how changing the layout of a hospital
can improve the health outcomes of patients include (Centre for Health Design, 2004):

Infection rate decline: when the Bronson Methodist Hospital incorporated private rooms and specially
located sinks, the hospital-acquired infections rate declined by 11%;

Medical errors minimised: medical errors at the Barbara Ann Karmanos Cancer Institute were
reduced by 30% after extra space was allocated for pharmaceutics, medical supplies were re-organised,
and panels were installed to reduce noise levels; and

Patient injuries decrease: the incidence of patient falls decreased by 75% in the intensive care unit at
Methodist Hospital when the location of nursing stations were moved to be in closer proximity to
patients’ rooms.
The BMA (2011) found that badly designed facilities can cause "anxiety, delirium, high blood pressure
and increased use of painkillers." Hospitals with little or no regard for their layout often create and
heighten stress levels for patients, staff, and visitors. This impact can range from psychological to
physiological and behavioural changes (Zimring, Joseph & Choudhary, 2004). The evidence shows that
the healthcare environment also has an effect on staff morale, efficiency, and effectiveness (Barach &
Dickerman, 2007).
Healthcare is a labour-intensive industry and a big proportion of that labour involves high skill levels
and thus also high salaries. 60 to 75% of hospital expenses can be attributed to labour expenses. Thus
a layout that promotes efficiency and minimises the strain of staff can have a significant effect on the
sustainability of a hospital (Carr, 2011). The negative effects of bad layout design can also be seen in
terms of travel time. According to Davis (2010), nurses on average spend time walking back and forth
one third of every shift due to badly designed layouts. Since maintenance and operations adds up to
80% of the total cost of a hospital over a life cycle of approximately 50 years, it is essential that facility
layout designers reduce the total life-cycle cost as much as possible through correct
design (Carr, 2011).
Before one can start designing the layout of a hospital it is important to fully understand what
operations a hospital entails, its characteristics, and how they differ from other facilities. Hospitals and
other healthcare facilities are characterised by extensive customer contact, a wide variety of providers,
and literally the life or death of patients as potential outcomes. The standard definition of a hospital is
“a facility whose staff provides services relating to observation, diagnosis, and treatment to cure or lessen
the suffering of patients” (Jacobs et al., 2010). Observation in this context refers to studying patients
and conducting various tests to arrive at a diagnosis. A diagnosis is a medical expert’s explanation of
the cause of the symptoms. Treatment is the course of action to be followed based upon the diagnosis.
All of the services provided in a hospital are usually organised around one or more of these three
areas.
3
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
There are several characteristics that set the operations of hospitals apart from other organisations.
These include (Jacobs et al., 2010):

Key personnel: the key operators are highly trained medical specialists who generate requests for
service (orders) but are also involved in delivering the service;

Process evaluation: the relationship between charging money and the actual performance is not as
direct as in production environments. The measures of quality and service are largely subjective;

Performance evaluation: hospitals do not have a simple line of command, instead there exists a
delicate balance of power between management, medical specialists, nursing staff, and referring doctors
and as such each party has different targets to measure performance;

Product specifications: product specifications in hospitals are often vague and subjective as opposed to
the complete and explicit specifications of end product requirements and delivery requirements in other
industries (i.e. for a hospital the patient is the product); and

Service oriented nature: hospital care cannot be stocked like commodities. It is a resource-oriented
service organisation.
What further complicates the design of a hospital layout is the various types of flow that exist. It is not
just the end product flow (in this case the treated patient) that is analysed; other flows include staff,
specimens, supplies, test results, and waste flow. There are also different users and stakeholders such
as patients, visitors, support staff, volunteers, suppliers, and owners (Carr, 2011). With so many design
considerations and different role players, the problem of designing a new hospital can be a daunting
task. The question thus becomes one of where does one start when designing the layout of a hospital.
The following questions are relevant:

What is the capacity of the hospital?

Which departments should be included in the hospital layout?

What is the capacity of each department?

Which rooms are necessary for each department?

How much space is required for each room?

Where should each department be located?

Which departments should and should not be placed in close proximity to one another?
Furthermore, the type of community plays a role on the hospital’s design. According to Kenny and
Duckett (2003), policies do not acknowledge or address the differences between rural and urban
healthcare facilities. There are many differences between the two that arguably promote that different
approaches should be taken. For instance, nurses in rural communities tend to fulfil a role that
requires them to have multiple skills and comprehensive knowledge, but are often not equipped or
educated for such a role. In these healthcare facilities medical care are often delivered without the
presence of the necessary healthcare professionals, especially doctors. This results in nurses having to
care for patients exhibiting more complicated problems (Eygelaar & Stellenberg, 2012). It is apparent
that rural communities differ from urban communities with regards to the challenges that they face
and that a differentiated hospital layout may serve the needs of these communities more adequately.
4
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
1.2.2 FACILITY LAYOUT DESIGN LITERATURE
There exist two generally accepted fields of study relating to hospital layout design, including the two
interrelated fields of quantitative and qualitative methods. The most regularly used quantitative
methods include layout models such as the Quadratic Assignment Problem (QAP), Quadratic Set
Covering Problem (QSCP), Linear Integer Programming Problem, Mixed Integer Programming
Problem (MIPP), Graph Theoretic Problem (GTP), Linear Continuous Model (LCM), and Linear Mixed
Integer Model (LMIM). When finding the optimal solution via these models, using Branch and Bound
Algorithms or Cutting Plane Algorithms is commonplace. Alternatively, most commonly when the
problem is too large, the model can be approximated using metaheuristics (e.g. Simulated Annealing)
or hybrid metaheuristics.
The qualitative approaches on the other hand include layout procedures such as Apple’s Plant Layout
Procedure, Muther’s Systematic Layout Procedure, and Reed’s Layout Procedure. There are also
various design considerations that should be incorporated for each of these.
Though popular, these methods do not necessarily take into account other important factors that are
specific to rural settings. It can therefore be said that there exists a gap in literature in this regard.
Remote rural region hospitals have certain challenges, constraints, and needs. This uniqueness of the
rural setting makes general hospital design methods inadequate. Furthermore, the field of designing
rural hospitals is limited, arguably due to the low economic incentive of investors and the fact that
they are oft seen as being in the realm of non-profit or governmental organisations. Therefore there
exists a need for an optimal layout design framework tailored specifically to rural communities.
1.3 Problem definition
Hospitals are complex systems and the design of the layout thereof is not a straightforward task. It is
suggested that hospitals in rural areas have different constraints and needs than hospitals in cities and
towns. It is unlikely that all rural and urban communities share the same lifestyles, health statuses,
prominent illnesses and consequently medical needs. Thus a different approach may need to be taken
when designing the layout of such a hospital. There exist quantitative as well as qualitative layout
design methods and the question is one of how these can be integrated into a framework for a hospital
layout. There are also numerous laws, standards and other design considerations with regards to
hospital design that the designer may not necessarily have knowledge of.
1.4 Aim and objectives
The primary objective of this study is to develop a generic decision-support framework using both
quantitative and qualitative layout design methods that will guide the user to design a near optimal
layout for a rural hospital while taking into consideration the relevant laws and standards as well as
5
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
the health outcomes of the surrounding rural community. The goal is not to replace the design team,
but to provide valuable input that can be used as a starting point for the design phase of a rural
hospital layout. Figure 1 shows the steps and phases for designing a hospital layout. The shaded blocks
indicate the steps that are addressed in this study.
Specify medical services
required to satisfy health needs
Determine interrelationships
among all services
Determine the space
requirements for all services
Generate hospital plans
Evaluate alternative hospital
plans
Implement the hospital plan
Define health needs
Select a hospital plan
Maintain and adapt the hospital
plan
PHASE 1
PHASE 2
PHASE 3
FIGURE 1: HOSPITAL LAYOUT DESIGN STEPS AND PHASES, ADAPTED FROM TOMPKINS, WHITE, BOZER AND TANCHOCO (2010)
The following secondary research objectives, delineated according to chapter, are pursued in this
study.
Quantitative layout design methods chapter:

Identify popular quantitative layout models used to design facility layouts;

Analyse the various components of these layout models;

Identify and analyse popular methods for solving these models;

Investigate other design methods such as layout software; and

Compare both the layout models, and the solution methods with each other.
Qualitative layout design methods and other design considerations chapter:

Identify popular qualitative design methods used to plan hospital layouts;

Analyse the various components of these qualitative design methods;

Identify design considerations for hospitals; and

Link the hospital design considerations with the quantitative methods.
Rural vs urban hospitals chapter:

Determine the commonalities among rural and urban hospitals;

Determine how laws and standards affect the design of a hospital;

Identify minimum dimensions and other criteria applicable to the design of a hospital layout;

Determine rural-specific constraints; and

Determine the layout implications of the rural-specific constraints.
Proposed rural hospital design framework chapter:

Describe the real world problem and how it differs from the research literature;
6
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION

Select the most adequate layout model and solution method;

Incorporate qualitative design methods and hospital design considerations into the framework;

Make necessary adjustments to the framework to accurately model the real world problem; and

Develop and present the framework.
Validation and verification chapter:

Determine whether the framework and findings of this study can be used to design a hospital layout.
Conclusions and recommendations chapter:

Summarise the findings of this study;

Describe the contributions of this study; and

Make recommendations for future research.
1.5 Research design
Primary data
Empirical
Ethnographic designs, surveys,
evaluation research, experiments,
field experiments, participatory
research, comparative studies,
Methodological studies
Discourse analysis, conversational
analysis, life history methodology
Conceptual studies, philosophical
analyses, theory and model building
Non-empirical
Secondary data analysis, modelling
and simulation studies, historical
studies, content analysis, textual
studies
Existing data
FIGURE 2: RESEARCH DESIGN MAPPING, ADAPTED FROM MOUTON (2011)
The research design method employed in this study is a non-empirical analysis of existing data
(a conceptual study), as shown in Figure 2. A range of secondary data (including articles, books,
scientific journals, and websites) is used in this study to perform a meta-analysis of existing
mathematical formulations regarding the FLP. The data is analysed so as to select the most adequate
model to design a hospital layout and make appropriate adjustments as necessary. The data also
includes South African laws and standards that add basic constraints to the problem of this study.
Furthermore, interviews were used in gathering expert opinions on adjustments, improvement
recommendations, and the validity of the framework. The nature of this study is that of operations
research, for which reason a case study is conducted to root it in reality.
1.6 Research methodology
7
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
Most literature for layout design problems can be classified into procedural and algorithmic
approaches (Shahin & Poormostafa, 2011). Algorithmic approaches involve quantitative methods
which usually simplifies design objectives and constraints with the aim of arriving at a substitute
(surrogate) objective function, through which the solution can more easily be obtained. According to
Tompkins, White, Bozer and Tanchoco (2010) most of the existing research take algorithmic
approaches. These algorithms can generate layout alternatives efficiently, but the results often fail to
capture all of the design objectives (Ahmad, 2005). Procedural approaches on the other hand are
qualitative methods consisting of several sequential steps but often include quantitative components.
These approaches rely on the experience of the designer and the results may consequently be highly
subjective, inefficient, and inferior in nature due to their lack of a sound scientific foundation. Thus,
neither
of
these
two
approaches
are
necessarily
effective
in
solving
real
world
problems (Ahmad, 2005). Therefore, the research methodology employed in this study is mixed
methods research which involves combining quantitative and qualitative approaches. It is still in its
adolescence and thus relatively unknown to many researchers (Leech & Onwuegbuzie, 2009).
Mixed methods research is defined by Creswell, Clark and Garrett (2008) as “both a method and
methodology for conducting research that involves collecting, analysing, and integrating quantitative
and qualitative research in a single study or a longitudinal program of inquiry”. Mixed methods utilise
the strengths of both quantitative and qualitative research. More insight can be gained from the
combination of both research types than by focusing on either form by itself (Creswell, 2015). Mixed
methods can be classified into concurrent and sequential designs (Dowbor & Zerger, 2014).
Concurrent designs refer to the collection of quantitative and qualitative research simultaneously,
whereas in sequential designs one type is collected after the other, classified as sequential explanatory
or sequential exploratory. In sequential designs, the emphasis is placed on the data that is collected
first and the secondary data is used to explain the first results. The emphasis can also be placed on
both quantitative and qualitative research equally, known as a convergent design.
According to Cresswell and Clark (2009) concurrent mixed methods are “employed to validate one
form of data with the other form, to transform the data for comparison, or to address different types of
questions”. Quantitative and qualitative research are conducted, analysed separately, and afterwards
compared and (or) combined. This is an ideal approach if it is unknown whether and how the two
research types can be integrated before conducting an in depth literature study. Thus, a concurrent
mixed methods approach is taken in this study. Figure 3 shows an outline of the research methodology
followed in this study. At first equal importance is placed on both quantitative and qualitative research
and they are analysed separately and in parallel. After this analyses, the method of integration
(embedding, connecting or merging) is determined, and as such the discussion of the mixed methods
approach is only addressed again in Section 5.2.
8
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
Problem definition
Hospital layout design
Quantitative
research
Qualitative
research
Literature review
Literature review
Quantitative layout design methods
Qualitative layout design methods
Hospital design considerations
Rural design considerations
Integration of quantitative and qualitative research
Solution formulation
Mathematical formulation of the
real-world problem
Development of generic rural
hospital design framework
Generic rural hospital design
framework
Validation and verification
Expert analysis
Application to case study
Final design
FIGURE 3: RESEARCH METHODOLOGY
1.7 Research scope and nature
There exist various types of hospitals, namely district hospitals (level one care), regional hospitals
(level two care), and provincial tertiary hospitals (level three care), central hospitals (level four care),
and specialised hospitals (level four care) (Conradie & Steyn, 2014). District hospitals normally
receive patients from clinics or health care centres and if a patient cannot be helped, they will be
referred to a regional hospital. Regional hospitals (and hospitals above level one care) are typically
only located in urban areas. Therefore district hospitals which can provide the basic level of care are
placed in rural communities. According to Van der Schyf and Flemming (2015) a district hospital is the
base hospital for a health district and is best suited to rural areas. For this reason it is assumed that the
hospital layout of this study will be developed for a district hospital.
In the case of a rural hospital, the available site space is usually not as restrictive as with typical urban
sites. Due to the relative difficulty of multi-floor building construction in rural regions where resources
are scarce and geographies remote, it seems logical to assume that a single floor layout should be
sufficient for a rural hospital. Without a low likelihood of being able to install an elevator it is safe to
assume that vertical travel between floors would only act so as to increase the complexity and
resources necessary for effective functioning. The number and location of vertical handling devices,
the resulting congestion and delays, added costs (both monetary and in terms of risk, including the
additional safety measures that will need to be put in place), and the possible lack of coordination
9
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
between departments on separate floors are important factors that detract from the appeal of a multifloor layout. For the abovementioned reasons, the scope of this study will be limited to single-floor
layouts.
The output of the developed framework will be designed to provide a point of departure for the
architectural concept design process, and as such will not be involved with detailing the precise
location of equipment, storage units, or beds. In other words the study will focus on designing a block
layout model which shows the locations and dimensions of the planning departments. Since there are
many rooms in each department of a hospital and there are many departments in one hospital, this
study will focus on finding a near-optimal arrangement of hospital departments, however not the
arrangement of rooms per se. Only the sizes of the rooms will be taken into consideration.
Lastly, due to the often remote and unbounded (unrestricted) nature of rural geographies, it is
assumed that the shape and space of the hospital site used in the development of the layout is
unrestricted, being determined by the preferences of the user of the framework.
1.8 Document outline
The document chapters are ordered to reflect the logical course of the study, as shown in Figure 4.
CHAPTER 2
CHAPTER 5
Identify and analyse quantitative
layout design methods
CHAPTER 3
Identify and analyse qualitative
layout design methods & design
considerations
CHAPTER 4
CHAPTER 6
Develop generic
hospital layout
design
framework
Identify hospital constraints
Validate and verify model

Expert analyses

Apply to case study
FINAL FRAMEWORK
Identify rural-specific constraints
and objectives
FIGURE 4: OUTLINE OF THESIS CHAPTERS
Each chapter has an introductory paragraph, and is concluded with a summary. The outline of the
chapters is now given.
Chapter 2 Quantitative layout design methods
The aim of this chapter is to provide an overview of the existing literature on quantitative layout
design methods. The problem of designing a layout is thoroughly discussed and the more popular
ways of mathematically modelling it are analysed along with its components. Various ways to solve
these models are also analysed. An overview on currently available quantitative layout software is also
provided. Lastly, the layout models, and solution methods are compared with each other.
10
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
Chapter 3 Qualitative layout design methods and hospital design considerations
This chapter investigates the designing of a hospital layout from a qualitative approach. Firstly, layout
types are explored and how hospital layouts are classified is discussed accordingly. Qualitative layout
methods are analysed, compared to the studied quantitative methods, and the best method is selected
and applied to the context of a hospital. This includes an analysis of each hospital department and the
functional relationships between them. The main design considerations of hospitals are described and
the relationships with the quantitative methods are established.
Chapter 4 Rural versus urban hospitals
In order to design a rural hospital layout correctly, the focus of this chapter is to determine
commonalities among rural and urban hospitals including the standards and laws that they need to
adhere to, e.g. minimum room dimensions and associated criteria. Thereafter, this chapter determines
rural-specific constraints and how they affect the design of a hospital layout.
Chapter 5 Proposed rural hospital design framework
This chapter combines the key findings from the previous chapters to develop a hospital layout design
framework for rural areas. The most adequate layout model is selected based on how well it models
the real world problem according to a set of criteria, the best solution method is chosen, and necessary
adjustments are then made. The qualitative design methods and hospital design criteria are then
integrated with the quantitative methods. Lastly, the development of a user-friendly interface for the
framework is fully described and how it is used is then explained.
Chapter 6 Validation and verification
The aim of this chapter is to verify and validate the framework and the key findings of this study. A
semi-structured interview process with experts is conducted and the necessary modifications are
made before applying it to a case study.
Chapter 7 Conclusions and recommendations
In ending, the chapter summarises the research conducted and concludes with the contribution of this
study, in addition to suggestions for future research.
1.9 Conclusion
This chapter introduced the research by first providing a background to the designing of hospital
layouts then setting out the research methodology that was followed in order to reach the research
objectives and aims.
It was found that the problem of designing a hospital layout is not a straightforward task and requires
various considerations, including laws and standards, different flows, patient centeredness, staffing
efficiency, and the type of community. A hospital’s layout plays an important role its ability to deliver
11
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 1 | INTRODUCTION
quality care to patients. A direct link was found between healthcare design and the medical outcomes
of patients. Furthermore, a hospital’s layout can have a substantially positive impact on hospital
expenses if its design promotes efficiency and minimising staffing needs.
Two commonly accepted fields that relate to the design of hospital layouts were introduced, namely
quantitative layout design methods and qualitative layout design methods. It was established that a
concurrent mixed methods approach is a suitable research methodology for this study. It may be
possible to integrate the quantitative and qualitative research methods which will be determined after
conducting an in depth literature study on both methods.
It was suggested that hospitals in rural areas have different constraints and needs than hospitals in
cities and towns. Thus a different approach may need to be taken when designing the layout of such a
hospital.
With a foundation in place regarding the approach and aim of the study, it is now necessary to follow
the roadmap developed in this chapter and comprehend the quantitative layout design methods found
in literature for designing a hospital.
12
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2
QUANTITATIVE LAYOUT DESIGN METHODS
“Not enough of our society is trained how to understand and interpret quantitative information. This
activity is a centrepiece of science literacy to which we should all strive—the future health, wealth, and
security of our democracy depend on it. Until that is achieved, we are at risk of making under-informed
decisions that affect ourselves, our communities, our country, and even the world.”
Tyso (1994)
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
13
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.1 Introduction
This chapter contains a review of the literature on topics related to the design of facilities, otherwise
known as the Facility Layout Problem (FLP). The focus of this chapter is to provide the reader with a
clear understanding of how one may design the layout of a facility using purely quantitative methods.
The chapter starts with a background on the field of operations research. Next, the FLP is discussed
and an overview of existing quantitative approaches commonly used to solve it is given. The more
popular layout models and solution methods are then analysed. Software available for designing
layouts is also discussed. The various quantitative methods addressed in this chapter are then
compared and the best method is proposed. Lastly, a chapter summary is given.
2.2 Operations Research
Winston and Goldberg (2004) define operations research as “a scientific approach to decision making
that seeks to best design and operate a system, usually under conditions requiring the allocation of scarce
resources.” The system in this context refers to an organisation of components that are interdependent
and work together towards the system’s goal (Winston & Goldberg, 2004). For example, Mercedes
Benz Company is a system that produces quality vehicles for the purpose of maximising its profit.
Operations research originated during the Second World War, wherein a group of scientists were
tasked with examining the strategies and tactics of various military operations so as to more efficiently
allocate scare resources during the war effort. As a result of this research on rendering military
operations as effective as possible, the term ‘operations research’ was coined (Winston &
Goldberg, 2004). Since the war involved complicated strategic and tactical problems, one cannot
expect adequate solutions from individuals in a single discipline. Therefore, groups of specialists in
engineering, economics, mathematics, statistics, and physical science were formed as units to jointly
deal with these problems and military operations (Agarwal & Srivastva, 2009). Due to the success of
these teams, various groups in the United States formed to study military logistical problems, the
effectiveness of aircraft flight patterns, the effective utilisations of electronic equipment of warfare,
and the planning of sea mining activities. After the war, the degree of success of the research teams
attracted the attention of industrial engineers to apply similar optimisation techniques to the fields of
business and industry. Since then, the scope of modern operations research has vastly increased,
spilling over into areas such as effective system utilisation and decision making within business and
industry, and even further into government and society as a whole.
One discipline that has more recently begun to benefit significantly from the field of operations
research, totally opposed to war yet inherently connected to it, is that of healthcare. It is especially
helpful in this field since it deals with the optimal usage of limited resources, which can mean the
difference between life and death for patients, be they in urban or rural geographies (Operations
14
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
Research Society of South Africa, 2014). Operations research highlights the trade-offs inherent in
determining the best solutions as well as provide managers with an informed basis for decision
making (Romeijn & Zenios, 2008).
2.3 Facility Layout Problem
The Facility Layout Problem (FLP) is a combinatorial optimisation problem that is ever-present in
manufacturing, service, and communication industries. It involves the optimal allocation of a set of
facilities to locations while conforming to area constraints, shape restrictions and relationships
between facilities (Tuzkaya, Gülsün, Tuzkaya, Onut & Bildik, 2013). Another description of the FLP is a
means of deciding upon the physical arrangement of a system (Meller & Gau, 1996). FLPs have
historically been applied to a large variety of problems, including the allocation of personnel and
equipment or the arrangement of buildings and sites (Liggett, 2000). A layout problem can either be
used for the arrangement of units in a new building or the relocation of units in an existing building.
During the conceptual design stage of a new building, it is worthwhile to test alternative arrangements
of spaces which can be used to determine other variables such as the ideal number of floors or the
required site space. Existing layouts can relocate units within the building in order to optimise the use
of spaces, e.g. the needs of a layout with regards to space often changes with time (Montreuil &
Laforge, 1992).
The FLP seeks the most efficient arrangement of a number 𝑛 of indivisible facilities with either equal
or unequal area requirements (Liggett, 2000). The FLP’s objective is typically to minimise material
handling costs (Tuzkaya et al., 2013). Other possible objectives include minimising total cost, travel
distance, flows, or travel time. It can also be rearranged into a maximisation problem, such as the
Koopmans and Beckmann formulation of the Quadratic Assignment Problem (QAP) which aims to
maximise the total net revenue, or conversely the Graph Theoretic Problem (GTP) which aims to
maximise a closeness rating measure. In Chapter 4 (Sections 4.3 and 4.4), exactly what these objectives
should be in the context of a rural hospital will be determined.
The FLP is usually subjected to two main constraints: the first involves space requirements and the
second facility restrictions such as fixed locations, empty locations or placing units within the available
space (Meller & Gau, 1996). Facility locations are generally permanent and therefore the optimal
setting is crucial for long term success due to the high impact on operational and logistical decisions.
The designed layout must not only perform well in the near future but also in a long term sense so as
to ensure lasting profitability. Factors influencing these decisions are largely time dependent, such as
but not limited to, environmental factors, population shifts, and evolution of market trends. Robust
facility designing is therefore an arduous task which requires much foresight in order to counter the
likely changes in the uncertain future (Wolf, 2011). Kumar and Suresh (2009) believe that the key to a
good layout lies in balancing and integrating the needs of people (personnel and customers), materials
15
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
(raw, finished, and in progress) and machinery in a manner which optimises the entire system,
however not necessarily the optimal solution for each of the individual needs.
2.3.1 QUANTITATIVE METHODS FOR MODELLING AND SOLVING THE FACILITY LAYOUT
PROBLEM
There exist many different ways of modelling and solving the FLP. In this chapter, various layout
models and their corresponding solution methods are examined. Each layout model has a certain set of
objective(s), assumptions, inputs and outputs and can either be solved optimally, e.g. with exact
methods or the optimal solution approximated, e.g. using metaheuristics. An outline of the FLP is
provided in Figure 5.




MODELLING
SOLVING
Layout models
Exact methods
Solve layout model optimally
Objectives
Assumptions
Inputs
Outputs
Metaheuristics
Approximate optimal solution
FIGURE 5: MODELLING AND SOLVING METHODS FOR THE FACILITY LAYOUT PROBLEM
A summary of layout models and solution methods found in literature that are used to solve the FLP
are shown in Table 1. Only reasonably cited articles found on Google Scholar and Web of Science were
included in this analysis. This analysis also includes layout software i.e. computer-based layout
algorithms which can be used to solve (and often model) the FLP.
TABLE 1: QUANTITATIVE METHODS USED TO MODEL AND SOLVE THE FACILITY LAYOUT PROBLEM
Author
Bazaraa and Goode
(1971)
Objective
Minimise material
handling cost
Bazaraa (1975)
Minimise material
handling cost
Maximise a
closeness rating
measure
Minimise
transportation
costs
Giffin and Foulds
(1987)
Heragu and Kusiak
(1987)
Model
QAP, QSCP
Solution method
FORTRAN computer program, Cutting Plane
Algorithm that exclude both integer and
non-integer solutions are generated at each
iteration
Branch and Bound Algorithm
GTP
Shortest Path Heuristic
QAP, QSCP, Linear
Integer
Programming
Problem, Mixed
Integer
Programming
Problem (MIPP),
GTP
Branch and Bound Algorithm, Cutting Plane
Algorithm, HC66, ALDEP, CORELAP, MAT,
PLANET, FATE, CRAFT, Revised Hiller
Algorithm, H63, FRAT, COFAD, Hybrid
Algorithms
QSCP
16
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
QUANTITATIVE METHODS USED TO MODEL AND SOLVE THE FACILITY LAYOUT PROBLEM (CONTINUED)
O'Kelly (1987)
Smith and MacLeod
(1988)
Montreuil (1991)
Heragu and Kusiak
(1991)
Leung (1992)
Minimise
transportation
costs
Minimise material
handling costs
Minimise material
handling costs
Minimise material
handling costs
Caccetta and
Kusumah (2001)
Maximise a
closeness rating
measure
Minimise financial
costs, maximise
operational
efficiency
Minimise material
handling costs
Minimise traveling
distance
Minimise material
handling cost
Minimise material
handling cost
Singh and Sharma
(2006)
Minimise material
handling cost
Arora and Saxena
(2007)
Minimise material
handling cost
Drira, Pierreval and
Hajri-Gabouj
(2007)
Minimise material
handling cost
Bianchi, Dorigo,
Gambardella and
Gutjahr (2009)
Minimise material
handling cost
Gülsün, Tuzkaya,
Onut and Bildik
(2013)
Minimize material
handling costs,
maximize
adjacency
requirements
Bland and Dawson
(1994)
Meller and Gau
(1996)
Meller, Narayanan
and Vance (1998)
Liggett (2000)
Quadratic Integer
Program
Enumeration Heuristic
QAP, QSCP
MIPP
Branch and Bound Algorithm
Linear Continuous
Model (LCM),
Linear Mixed
Integer model
(LMIM), MIPP
GTP
Powell method of conjugate direction used
with the penalty method
QAP
Tabu Search (TS), Simulated Annealing
(SA), Hybrid of TS and SA gave superior
results for large-scale layout problems
QAP, GTP, MIPP
MATCH, SPIRAL, CRAFT, LOGIC, MULTIPLE,
FLEX-BAY
Branch and Bound Algorithm
QAP, MIPP
QAP
QAP, QSCP, Linear
Integer
Programming
Problem, MIPP,
GTP
QAP, GTP, MIPP
QSCP, Linear
Programming
Problem, QAP
QAP, MIPP
Stochastic
Combinatorial
Optimisation
Problems
QAP
17
CRAFT, MULTIPLE, SA, Genetic Algorithm
(GA), Hybrid approaches
GTP based heuristics such as Deltahedron
Method, Constructive Heuristic, Wheel
Expansion Algorithm, Kim-Kim Algorithm,
Tessa, Green-Al Hakim Algorithm,
Branch and Bound Algorithm used for small
instances, MATCH, SPIRAL, CRAFT, LOGIC,
MULTIPLE, FLEX-BAY, SA, GA
Branch and Bound Algorithm
Branch and Bound Algorithm, CORELAP,
ALDEP, COFAD, SHAPE, CRAFT, FRAT,
DISCON, TS, SA, GA, Hybrid methods, Ant
Colony Optimisation
Ant Colony Optimisation, Evolutionary
Computation, SA, TS
GA, SA, Hybrid approach (HGASA), SA
performed best in terms of fitness and time
requirements
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
QUANTITATIVE METHODS USED TO MODEL AND SOLVE THE FACILITY LAYOUT PROBLEM (CONTINUED)
Büyüksaatçi and
Baray (2014)
Mmise total cost
(overall efficiency
of operations and
reduces total
operating
expenses)
Single Row Facility
Layout Problem
Bacterial Foraging Optimisation
Algorithm, Firefly Algorithm, Bat Algorithm,
Particle Swarm Optimisation
Modelling the Facility Layout Problem
Table 1 supports the fact that most of the methods aim to optimise the material handling cost as the
main objective of the FLP. Other objectives include minimising total cost, transportation cost and
maximising a closeness rating measure. According to Tuzkaya et al. (2013) most of the FLPs are based
on the QAP, LIPP, MIPP and GTP. From Table 1 it is clear that the QSCP is also a prominent layout for
modelling the FLP. Two new layout models are developed by Heragu and Kusiak (1991), namely the
Linear Continuous Model (LCM) and the Linear Mixed Integer Model (LMIM) (Asl & Wong, 2015). The
analysis in Table 1 indicated that the following five layout models are more popular, namely the QAP
(included 11 times), MIPP (included 7 times), GTP (included 4 times), QSCP (included 3 times), and
LIPP (included 2 times). Therefore, these five layout models along with the relatively less studied LCM
and LMIM are chosen to be further discussed in this study. Some of these models have variations, e.g.
the Koopmans and Beckmann formulation and the Trace formulation of the QAP (further explained in
Section 2.4.2).
Solving the Facility Layout Problem
Compared to modelling the FLP, there are many more methods available for solving the FLP. The
reason for this could be that it is not straightforward to solve the FLP. Bozer, Meller and
Erlebacher (1991) argue that there are two key reasons for this. Firstly, there are no generally
accepted objective functions which capture all the relevant aspects of the problem. Secondly, with
commonly accepted functions, finding the optimal solution is near-impossible since it often involves
large-scale problems. The run-time for solving these layout models usually increases dramatically with
the problem size and often only small sized instances can be solved practically. For example, consider
the QAP formulation which involves assigning 𝑛 facilities to 𝑛 locations in such a way that each facility
has one location and each location has only one facility. If 10 facilities are to be assigned to 10
locations, then n! (equal to 3,628,800) permutations has to be checked which requires a large amount
of computation time (Bhati & Rasool, 2014).
A timeline of solution methods used in literature to solve the FLP is shown in Figure 6. Solution
methods for the FLP may be divided into two categories, namely exact methods, and metaheuristics.
Exact methods are able to find the optimal solution to a problem. Examples include the Branch and
Bound Algorithm and the Cutting Plane Algorithm. As just explained, the FLP is very difficult to solve
for large problems and only FLPs with small problem sizes can be solved optimally using exact
18
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
methods (Meller, Narayanan & Vance, 1998). It was found that exact methods cannot be used to solve
the FLP for problem sizes more than 15 to 20 facilities (Tuzkaya et al., 2013).
1957
1957
Mathematical
1963/01/01programming by Koopmans and Beckmann (1957)
CORELAP by Lee (1963)
1964/01/01
CRAFT by Buffa (1964)
1967/01/01
ALDEP by Seehof and Evans (1967)
1986/01/01
SHAPE by Hassan, Hogg and Smith (1986)
1987/01/01
MATCH by Montreuil, Ratliff and Goetschalckx (1987)
1992/01/01
LOGIC by Tam (1992)
1992
Simulated annealing for row layout by Heragu and Alfa (1992)
1993/01/01
Simulated annealing for QAP by Suresh and Sahu (1993)
1994/01/01
MULTIPLE by Bozer, Meller and Erlebacher (1994)
1994/01/01
Tabu Search for QAP by Skorin-Kapov (1994)
1994/01/01
Genetic Algorithm for dynamic QAP by Conway and Venkataramanan (1994)
1997/01/01
Branch and Bound by Lacksonan (1997)
1998/01/01
Genetic algorithm by Islier (1998)
2000
Tabu-search based by Helm and Hadley (2000)
2000
Hybrid approaches by Liggett (2000)
2001
Graph Theoretic based heuristics by Cacetta and Kusumah (2001)
2003/01/01
Extended distance based by Castillo and Peters (2003)
2007/01/01
Branch and Bound by Arora and Saxena (2007)
2009/01/01
Ant Colony Optimisation by Bianchi et al. (2009)
2013/01/01
Hybrid of SA and GA by Tuzkava et al. (2013)
2014/01/01
Firefly algorithm by Büyüksaatçi and Baray (2014)
2016
FIGURE 6: TIMELINE OF QUANTITATIVE METHODS USED TO SOLVE THE FACILITY LAYOUT PROBLEM
In order to approximate optimal solutions for large and complex problems, metaheuristics and
heuristics were developed to find near optimal solutions within a reasonable computation time.
Commonly used methods for solving the FLP are Ant Colony Optimisation (ACO), Genetic Algorithms,
Local Search (LS), Particle Swarm Optimisation (PSO), Simulated Annealing (SA), and Tabu Search
(TS) (Bhati & Rasool, 2014; Büyüksaatçi & Baray, 2014). These more popular methods are chosen to
be further analysed in this study (Section 2.5.2). The combination of metaheuristics, known as hybrid
metaheuristics, can also be used for the purpose of finding a superior layout solution and is therefore
discussed in more detail in Section 2.5.3.







MODELLING
SOLVING
Layout models
Exact methods


Quadratic Assignment Problem
Quadratic Set Covering
Problem
Linear Integer Programming
Problem
Mixed Integer Programming
Problem
Graph Theoretic Problem
Linear Continuous Model
Linear Mixed Integer Model
Branch and Bound Algorithm
Cutting Plane Algorithm
Metaheuristics







Local Search method
Simulated Annealing
Tabu Search
Genetic Algorithms
Ant Colony Optimisation
Particle Swarm Optimisation
Hybrid metaheuristics
FIGURE 7: CHOSEN LAYOUT MODELS AND SOLUTION METHODS FOR FURTHER ANALYSIS
19
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
A summary of the chosen layout models and solution methods that are further analysed in this study
are shown in Figure 7.
2.3.2 QUANTITATIVE METHODS SOFTWARE PACKAGES
Another way to quantitatively model and solve the FLP is via layout software. Layout software uses
computer-based algorithms, often classified as either construction algorithms or improvement
algorithms. There are also algorithms which can be used as both. Figure 6 and Table 1 show timelines
of layout software found in literature. Examples of layout software include CRAFT, ALDEP, MATCH,
CORELAP, COFA, SPIRAL, MULTIPLE, RAFT, SHAPE, LOGIC, and FLEX-BAY. Table 1 indicates that
CRAFT (included 5 times), LOGIC (included 3 times) and MULTIPLE (included 2 times) are common
software packages. These methods are chosen to be further discussed in Section 2.6.
2.3.3 APPLICATION OF THE FACILITY LAYOUT PROBLEM TO HOSPITAL LAYOUTS
One of the first applications of the FLP to hospitals was by Elshafei (1977) when the hospital layout
was formulated as a QAP in order to minimise the efforts of patients walking within the hospital.
Elshafei focussed specifically on one unit of a large hospital, namely the out-patient department and
analysed the treatment of patients from 17 clinics within the department. In this example the task at
hand was to assign clinics to locations within the department. The problem of this study focuses on the
locations of departments within a hospital and from this point onwards, the FLP refers to the assigning
of departments within a hospital. Another study by Tobias (1986) used a GTP approach to design a
hospital layout and used CORELAP to find a solution. Hahn and Krarup (2001) reviewed solving the
QAP for large problems. More recently Feyzollahi, Shokouhi, Yazdi and Tarokh (2009) have designed a
hospital layout on the basis of the QAP model.
These studies found the application of the FLP to new or existing hospital layouts valuable. However,
very few studies were found in the literature and the focus was only on the GTP and QAP formulations.
Within the context of this study, the FLP refers to assigning 𝑛 departments to locations within a
hospital. Each location can house only one department and each department occupies one (or more
locations).
2.4 Layout models
Layout models are used to describe and formulate the FLP. The following layout models are discussed
in this section, namely the Quadratic Assignment Problem, Quadratic Set Covering Problem, Linear
Integer Programming Problem, Mixed Integer Problem, Graph Theoretic Problem, Linear Continuous
Model and the Linear Mixed Integer Model. For each of the models listed, the main objective(s),
accompanying notation, assumptions, inputs, outputs, and variations are analysed in the sections that
follow.
20
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.4.1 CLASSIFICATION OF MODELS
The abovementioned layout models can be classified as being either discrete or continuous. Discrete
layout models limit the locations of the departments to be placed at a number of predefined positions,
otherwise referred to as candidate sites. Continuous layout models allow new departments to be
located anywhere within the space being modelled. As a result of this the distances between each pair
of departments are continuous variables. This differs from discrete layout models in that the distance
between any pair of departments is based on the assignment of departments to defined locations. The
advantage of this discrete representation is its allowance for simple linear assignment constraints and
the way in which it easily avoids inter-centre overlap, which often poses difficulties for optimising the
continuous space (Montreuil, Brotherton & Marcotte, 2002).
TABLE 2 LAYOUT MODELS CLASSIFIED ACCORDING TO DISCRETE AND CONTINUOUS
Discrete
Quadratic Assignment Problem
Quadratic Set Covering Problem
Linear Integer Programming Problem
Continuous
Mixed Integer Programming Problem
Graph Theoretic Problem
Linear Continuous Model
Linear Mixed Integer Model
Table 2 shows how the seven layout models can be classified. Refer to the following research for more
explanation on the classification of each model: Adams (2010) for QAP and QSCP; Drira, Pierreval and
Hajri-Gabouj (2007) for QAP, MIPP and LMIM; Schenker, Kandel, Bunke and Last (2005) for GTM;
Heragy and Kusiak (1991) for LCM and LMIM. The MIPP Problem and the LMIM contain continuous as
well as discrete variables.
2.4.2 QUADRATIC ASSIGNMENT PROBLEM
The QAP is one of the most used methods for designing a facility layout. This being said, in practice
FLPs are in general more complex than the QAP formulation since the QAP considers only the problem
of assigning 𝑛 equal-sized departments to 𝑛 pre-determined locations (Meller, Narayanan &
Vance, 1998). This assignment occurs according to a cost function of distance and flow between
departments as well as the cost of placing a department at a specific location. Thus, the QAP acts so as
to allocate departments to locations in such a way that the total costs will be minimised (Burkard, Cela,
Pardalos & Pitsoulis, 1999). However, the objectives of the QAP can also be arranged to either
minimise material handling cost, travel time, travel distance or flows (Singh & Sharma, 2006).
According to Heragu and Kusiak (1987) the QAP has been frequently used to model the FLP, but it is
not able to serve as a model for all FLPs. The QAP model cannot be applied if how far each location is
from another is unknown, e.g. the machine layout problem cannot be solved as a QAP model since the
distances between locations depends on the arrangement of the machines. Layout problems on the
other hand with equal areas can be solved since the locations have equal distances and are thus
21
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
independent. Thus, the distances stay the same regardless of the department’s arrangement. The
general formulation of the QAP is given in Table 3.
TABLE 3: GENERAL FORMULATION OF THE QUADRATIC ASSIGNMENT PROBLEM (HERAGU & KUSIAK, 1987; KONAK, KULTUREL-KONAK, NORMAN &
SMITH, 2006)
Objective(s)
Notation
Assignment of 𝑛 departments to 𝑛 locations
Minimise total cost/traveling distance/flows/traveling time
𝑛
Minimise
𝑛
𝑛
𝑛
𝑛
𝑛
∑ ∑ ∑ ∑ 𝑓𝑖𝑘 𝑐𝑗𝑙 𝑥𝑖𝑗 𝑥𝑘𝑙 + ∑ ∑ 𝛾𝑖𝑗 𝑥𝑖𝑗
𝑖=1 𝑗=1 𝑘=1 𝑙=1
𝑛
Subject to
(1)
𝑖=1 𝑗=1
∑ 𝑥𝑖𝑗 = 1, 𝑖 = 1,2, … , 𝑛,
(2)
𝑗=1
𝑛
∑ 𝑥𝑖𝑗 = 1, 𝑗 = 1,2, … , 𝑛,
(3)
𝑖=1
𝑛
𝛾𝑖𝑗
𝑓𝑖𝑘
𝑐𝑗𝑙
𝑥𝑖𝑗
𝑥𝑘𝑙
=
=
=
=
=
=
𝑥𝑖𝑗 ∈ {0,1}, 𝑖, 𝑗 = 1,2, … , 𝑛.
(4)
total number of departments/locations
fixed cost of locating and operating department 𝑖 at location 𝑗
flow of patients between department 𝑖 and department 𝑘
cost of transporting one patient between location j and location 𝑙
1 if department 𝑖 is at location 𝑗; 0 otherwise
1 if department k is at location 𝑙; 0 otherwise
Assumptions
𝑡𝑖𝑗 includes gross revenues minus cost of primary input but does not include the
transportation cost of patients between departments
𝑓𝑖𝑘 is independent of the locations of the departments
𝑐𝑗𝑙 is independent of the departments and that it is cheaper to transport patients directly from
department 𝑖 to department 𝑘 than through a third location
The number of departments is equal to the number of locations
The available space can be divided into equal blocks
Equal sized departments
Inputs
The total number of departments or locations
The fixed cost of locating and operating department 𝑖 at location 𝑗
The flow of patients between department 𝑖 and department 𝑘
The cost of transporting one patient between location 𝑗 and location 𝑙
Outputs
Assignment of departments to locations
Minimum total cost/traveling distance/flows/traveling time
The QAP formulation is difficult to solve for even small problem sizes. Cela (1998) found that it can be
solved with reasonable limits up to a problem size of 20. As a result of this, other researchers have
simplified the QAP formulation. Two special cases of this include the Linear Assignment Problem
(LAP) and the Travelling Salesman Problem (TSP). A QAP (in Table 3) is reduced to a LAP when the
𝑓𝑖𝑘 ’s are either zero or identical. The new objective is shown in (5).
𝑛
Minimise
𝑛
∑ ∑ 𝑎𝑖𝑗 𝑥𝑖𝑗
𝑖=1 𝑗=1
22
(5)
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
𝑛
Subject to
(6)
∑ 𝑥𝑖𝑗 = 1, 𝑖 = 1, … , 𝑛,
𝑗=1
𝑛
(7)
∑ 𝑥𝑖𝑗 = 1, 𝑗 = 1, … , 𝑛,
𝑖=1
(8)
𝑥𝑖𝑗 ∈ {0,1}, 𝑖, 𝑗 = 1, … , 𝑛.
(6) to (8) suggest that if the 𝑥𝑖𝑗 ’s are shown in a matrix (𝑋 = [𝑥𝑖𝑗 ]), this matrix will be a permutation
matrix (meaning a matrix with zeros and one nonzero entry, i.e. 1, in each row and column) (Heragu &
Kusiak, 1987). An additional constraint is that the permutation matrix is cyclic, meaning that all the
elements in the permutation matrix are shifted by a fixed offset (elements are shifted off the end and
inserted back at the beginning) (Weisstein, 2015).
According to Cela (1998) the QAP is usually defined as in Table 3 since the problem formulation
expresses the combinatorial structure of the QAP better than alternative equivalent formulations.
However, the alternatives are useful when Subgradient Optimisation (a type of iterative method for
solving convex minimisation problems) or MIPP, and Semidefinite Programming (a subfield of convex
optimisation concerned with the optimisation of a linear objective function) are applied. There are two
alternative formulations of the QAP. They are the Koopmans and Beckmann formulation and the Trace
formulation, each of which are outlined now.
Koopmans and Beckmann formulation
The QAP was first formulated by Beckmann and Koopmans (1957) in order to model a plant location
problem. Since then the number of real-life problems which can be mathematically modelled by QAP
has steadily increased, spreading into a number of associated fields. A few examples of these
applications are placement problems, manufacturing, scheduling, very-large-scale integration design,
parallel and distributed computing, and statistical data analysis (Lawler, 1962). The Koopmans and
Beckmann formulation of the QAP is given in Table 4.
TABLE 4: KOOPMANS AND BECKMANN FORMULATION OF THE QUADRATIC ASSIGNMENT PROBLEM (BECKMANN & KOOPMANS, 1957; HERAGU &
KUSIAK, 1987)
Objectives
Notation
Assign 𝑛 departments to 𝑛 locations
Maximise the total net revenue
∑ 𝑝𝑘𝑖 𝑥𝑘𝑖 − ∑ ∑ 𝑐𝑖𝑗 𝑥𝑘𝑙,𝑖𝑗
Maximise
𝑘,𝑖
Subject to
𝑘,𝑙
𝑥𝑘𝑖 𝑢𝑘𝑙 + ∑ 𝑥𝑘𝑖 = 𝑥𝑙𝑖 𝑢𝑘𝑙 + ∑ 𝑥𝑘𝑖,𝑖𝑗 , 𝑘, 𝑙, 𝑖 = 1, … , 𝑛,
𝑖
(9)
𝑖,𝑗
(10)
𝑗
∑ 𝑥𝑘𝑖 = 1, 𝑘 = 1, … , 𝑛,
(11)
𝑖
∑ 𝑥𝑘𝑖 = 1, 𝑖 = 1, … , 𝑛,
𝑘
23
(12)
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
KOOPMANS AND BECKMANN FORMULATION OF THE QUADRATIC ASSIGNMENT PROBLEM (BECKMANN & KOOPMANS, 1957; HERAGU & KUSIAK, 1987)
(CONTINUED)
𝑥𝑘𝑖 ≥ 0, 𝑥𝑘𝑖,𝑖𝑗 ≥ 0, 𝑥𝑘𝑖,𝑖𝑖 = 0, 𝑘, 𝑙, 𝑖, 𝑗 = 1, … , 𝑛.
Notation
𝑛
𝑝𝑘𝑙
=
=
𝑢𝑘𝑙
=
𝑐𝑖𝑗
𝑥𝑘𝑙,𝑖𝑗
=
=
𝑥𝑘𝑖
=
(13)
Same formulation as in Table 3
Semi-net revenue from the operation of department 𝑘 at location
𝑖. In other words the gross revenue less cost of primary inputs,
but before subtracting transportation cost of intermediate
products between departments
Set of non-negative numbers that represent the required
commodity flows in weight units from department 𝑘 to
department 𝑙. 𝑘 ≠ 1. l = 1, …, 𝑛
Same formulation as in Table 3
1 if there is flow from location 𝑖 to location 𝑗 of the commodity
which is supplied by department 𝑘 to department 𝑙; 0 if there is
no flow
Same formulation as in Table 3
Assumptions
The flow coefficients 𝑢𝑘𝑙 are independent of the locations assigned
The transportation cost coefficients 𝑐𝑖𝑗 are assumed independent of the plant assignments
and applicable to all amounts and compositions of flows
The semi-net revenue from a given fractional plant 𝑘 in a given location 𝑖 are proportional to
the size 𝑥𝑘𝑖 of the fraction
The input and output flows are proportional to the size 𝑥𝑘𝑖 of the fraction
𝑝𝑘𝑙 includes gross revenues minus cost of primary input but does not include the
transportation cost of material between plants
Equal sized departments
Inputs
Total number of departments and locations
Semi-net revenue from the operation of department 𝑘 at location 𝑖
Set of nonnegative numbers that represent the required commodity flows in weight units
from plant 𝑘 to plant 𝑙
Set of positive numbers that represent the cost of transportation for the unit flow from
location 𝑖 to location 𝑗
Outputs
Assignment of 𝑛 departments to 𝑛 locations
Maximised total net revenue
Trace formulation
The Trace formulation uses linear algebra and the trace function to determine the lower bounds for a
cost objective function (Loiola, De Abreu, Boaventura-Netto, Hahn & Querido, 2007). The trace
function involves the summation of the main diagonal elements of a matrix. The trace formulation was
used in several works including Handley (1994), Karisch and Rendl (1995), and Anstreicher and
Brixius (2001). It can also be used as a tool for manipulating the algebraic problem data (Cela, 2013).
The trace formulation is shown in Table 5.
24
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
TABLE 5: TRACE FORMULATION OF THE QUADRATIC ASSIGNMENT PROBLEM (BURKARD ET AL., 2009; COMMANDER, 2003)
Objectives
Notation
Assignment of 𝑛 departments to 𝑛 locations
Minimise total cost/traveling distance/flows/traveling time
Minimise
Subject to
𝑡𝑟𝑎𝑐𝑒(𝐶𝑋𝐷𝑡 𝑋 𝑡 )
𝑋 ∈ 𝑋𝑛
(14)
(15)
𝑡𝑟𝑎𝑐𝑒(𝐴) = ∑ 𝑎𝑖𝑖
(16)
𝑛
𝑖=1
A
C
𝑛
D
=
=
=
=
X
=
𝑛 x 𝑛 matrix
Cost matrix. 𝐶 = [𝑐𝑖𝑗 ] where 𝑐𝑖𝑗 is defined the same as Table 3
Same formulation as in Table 3
Distance matrix. 𝐷 = [𝑑𝑘𝑙 ] where 𝑑𝑘𝑙 represents the distance
between location 𝑘 and location 𝑙
Permutation matrix. 𝑋 = [𝑥𝑖𝑗 ] an 𝑛 x 𝑛 matrix where 𝑥𝑖𝑗 is similar
as defined in Table 3
Assumptions
The trace of a square matrix is defined as the sum of its diagonal elements
Equal-sized departments
Inputs
Cost matrix
Distance matrix
Permutation matrix
Outputs
Assignment of 𝑛 departments to 𝑛 locations
Minimum total cost/traveling distance/flows/traveling time
One of the main disadvantages of the QAP formulation is assuming that the departments are equalsized since that is often difficult to justify in practical cases. According to Konak, Kulturel-Konak,
Norman and Smith (2006), it is possible to adjust the QAP formulation to include unequal departments
by using small equal area grids and forbidding separating the grids of the same department via the
assignment of very high flows between them. This allows departments to have various sizes and
rectangular shapes. However, this leads to a large number of integer variables which means that even
small problem sizes cannot be optimally solved. Since the high artificial flows enforce unintended
constraints between grids of the same department, it is likely to lead to a poor layout solution (Bozer &
Meller, 1997).
2.4.3 QUADRATIC SET COVERING PROBLEM
Bazaraa and Goode (1971) developed an improved formulation for Bellmore and Ratliff’s linear setcovering algorithm for the quadratic case. The new formulation enables the algorithm to solve
problems of the inequality type as well as problems of equality and mixed types by incorporating an
additional penalty term in the objective function. The QSCP divides the available floor space into equal
blocks (Heragu & Kusiak, 1987). Grid-points are formed which serve as the available locations for the
centres. The constraints of the QSCP formulation ensure that each department has one location and
each location is occupied by at most one department (Welgama & Gibson, 1995). Several candidate
locations are made available for each centre of a department which prevents each location from being
25
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
available to every centre and allows the accommodation of different centre sizes. This implies that the
number of combinations being evaluated is smaller. The QSCP is shown in Table 6.
TABLE 6: QUADRATIC SET COVERING PROBLEM (HERAGU & KUSIAK, 1987; BESTER, 2006)
Objectives
Assignment of 𝑛 departments to locations
Minimisation material-handling cost (as well as the relocation cost of centres)
Notation
𝑛
Minimise
𝑛
𝐼(𝑖) 𝐼(𝑘)
𝑛
𝐼(𝑖)
∑ ∑ ∑ ∑ 𝑓𝑖𝑘 𝑥𝑖𝑗 𝑥𝑘𝑙 𝑑𝑗𝑙 + ∑ ∑ 𝑡𝑖𝑗 𝑥𝑖𝑗
𝑖=1 𝑘=1 𝑗=1 𝑙=1
𝑖=1 𝑗=1
𝐼(𝑖)
Subject to
(17)
∑ 𝑥𝑖𝑗 = 1, 𝑖 = 1,2, … , 𝑛,
(18)
𝑗=1
𝑛
𝐼(𝑖)
∑ ∑ 𝑎𝑖𝑗𝑡 𝑥𝑖𝑗 ≤ 1, 𝑡 = 1,2, … , 𝑞,
(19)
𝑖=1 𝑗=1
𝑞
=
𝑑𝑗𝑙
=
𝑓𝑖𝑘
𝑡𝑖𝑗
𝐼(𝑖)
𝐽𝑖 (𝑗)
𝑥𝑖𝑗 , 𝑥𝑖𝑘
𝑎𝑖𝑗𝑡
=
=
=
=
=
=
𝑥𝑖𝑗 ∈ {0,1}, 𝑖 = 1,2, … , 𝑛; 𝑗 = 1,2, … , 𝐼(𝑖).
(20)
Number of blocks into which the total area occupied by all
departments is divided into
Distance between the centroids of locations 𝑗 and l if department 𝑖 is
assigned to location 𝑗 and department 𝑘 is assigned to location 𝑙
Similar to Table 3
Fixed cost of locating and operating department 𝑖 at location k
Set of candidate locations for department 𝑖
Set of blocks occupied by department 𝑖 if it is assigned to location 𝑗
Same definitions as in Table 3
1 if block 𝑎 ∈ 𝐽𝑖 (𝑗); 0 otherwise
Assumptions
The candidate locations for each centre have to be established in advance
The available space is divided into blocks of equal size
𝑡𝑖𝑗 includes gross revenues minus cost of primary input but does not include the
transportation cost of patients between departments
𝑓𝑖𝑘 is independent of the locations of the departments
The number of departments is equal to the number of locations
Inputs
Fixed cost of locating and operating department 𝑖 at location 𝑘
Flow of patients between department 𝑖 and department 𝑘
Distance between the centroids of locations 𝑗 and 𝑙 if f department 𝑖 is assigned to location 𝑗
and department 𝑘 is assigned to location 𝑙
Number of blocks into which the total area occupied by all departments is divided into
Set of candidate locations for department 𝑖
Outputs
Assign 𝑛 departments to locations
Set of blocks occupied by department 𝑖 if it is assigned to location 𝑗
Minimum material handling cost
One must note that 𝑑𝑗𝑙 refers to the rectilinear distance (e.g. |𝑥𝑖 − 𝑥𝑗 |+|𝑦𝑖 − 𝑦𝑗 |) between two locations
since the departments are laid out in the form of a rectangular grid and a straight line of travel is
unlikely. The objective function given in (17) consists of two terms that do not have the same units.
The first term represents flow between departments (measured in m3/s) multiplied by the distance
26
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
(measured in m), which results in a term with a unit of m 4/s. The second term is the monetary unit.
Since there is no known conversion rule between these units, Bester (2006) incorporated a
real-valued scaling parameter (0 < 𝜃 < 1) which compensates for this discrepancy. This parameter
may be varied manually in order to test different conversion rules. The new objective function is
shown in (21).
𝑛
Minimise
𝑛
𝐼(𝑖) 𝐼(𝑘)
𝑛
𝐼(𝑖)
(21)
𝜃 (∑ ∑ ∑ ∑ 𝑓𝑖𝑘 𝑥𝑖𝑗 𝑥𝑘𝑙 𝑑𝑗𝑙 ) + (1 − 𝜃) (∑ ∑ 𝑡𝑖𝑗 𝑥𝑖𝑗 )
𝑖=1 𝑘=1 𝑗=1 𝑙=1
𝑖=1 𝑗=1
A disadvantage of the QAP formulation is that by dividing the floor space into smaller blocks to
accommodate different sizes, it leads to an increase in the problem size thus making it harder to
solve (Bazaraa & Goode, 1971). Furthermore, the candidate solutions for each centre have to be
established in advance and the specification of centre shapes might not be practical in the sense that
some centres have L-shapes but the model assigns rectangular shapes to them.
2.4.4 LINEAR INTEGER PROGRAMMING PROBLEM
Linear Integer Programming Problems (LIPP) are optimisation problems that exhibit both linear
objective functions and linear inequality constraints (Suo, 2012). A wide variety of real life problems
can be formulated as a LIPP besides the FLP, for example, the knapsack-capital budgeting problem,
machine layout problem, maximum flow problems, weighted matching problems, and scheduling
problems (Genova & Guliashki, 2011). In such problems, the variables are constrained to be integers. It
is possible to transform some optimisation problems with non-linear objective functions and linear
constraints into a LIPP by approximating the non-linear function via piecewise linear functions. Refer
to Bertsekas (2003) for this transformation. The FLP was first modelled as a LIPP by Lawler (1962). In
his article he proves that the formulation that is shown in Table 7 is equivalent to the QAP. The QAP
has 𝑛2 variables and 2𝑛 constraints, while LIPP has a total of 𝑛4 + 2𝑛 + 1 constraints (Suo, 2012).
TABLE 7: LINEAR INTEGER PROGRAMMING PROBLEM (LAWLER, 1963; HERAGU & KUSIAK, 1987)
Objectives
Notation
Minimise total cost
Assignment of 𝑛 departments to 𝑛 locations
𝑛
Minimise
𝑛
𝑛
𝑛
∑ ∑ ∑ ∑ 𝑏𝑖𝑗𝑘𝑙 𝑦𝑖𝑗𝑘𝑙
(22)
𝑖=1 𝑗=1 𝑘=1 𝑙=1
𝑛
Subject to
∑ 𝑥𝑖𝑗 = 1, 𝑖 = 1,2, … , 𝑛,
(23)
𝑗=1
𝑛
∑ 𝑥𝑖𝑗 = 1, 𝑗 = 1,2, … , 𝑛,
𝑖=1
𝑛 𝑛
𝑛
(24)
𝑛
∑ ∑ ∑ ∑ 𝑦𝑖𝑗𝑘𝑙 = 𝑛2 ,
(25)
𝑖=1 𝑗=1 𝑘=1 𝑙=1
𝑥𝑖𝑗 + 𝑥𝑘𝑙 − 2𝑦𝑖𝑗𝑘𝑙 ≥ 0, 𝑖, 𝑗, 𝑘, 𝑙 = 1,2, … , 𝑛,
27
(26)
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
LINEAR INTEGER PROGRAMMING PROBLEM (LAWLER, 1963; HERAGU & KUSIAK, 1987) (CONTINUED)
𝑦𝑖𝑗𝑘𝑙 = 𝑥𝑖𝑗 . 𝑥𝑖𝑙
𝑥𝑖𝑗 ∈ {0,1}, 𝑖, 𝑗 = 1,2, … , 𝑛,
𝑦𝑖𝑗𝑘𝑙 ∈ {0,1}, 𝑖, 𝑗, 𝑘, 𝑙 = 1,2, … , 𝑛,
𝑓𝑖𝑘 𝑐𝑗𝑙 + 𝑡𝑖𝑗
𝑖 = 𝑘, 𝑗 = 1
𝑏𝑖𝑗𝑘𝑙 = {
𝑓𝑖𝑘 𝑐𝑗𝑙
𝑖 ≠ 𝑘 𝑜𝑟 𝑗 ≠ 𝑙
Notation
𝑥𝑖𝑗
𝑥𝑖𝑙
𝑐𝑗𝑙
𝑦𝑖𝑗𝑘𝑙
=
=
=
=
𝑓𝑖𝑘
𝑡𝑖𝑗
𝑏𝑖𝑗𝑘𝑙
𝑛
=
=
=
=
(27)
(28)
(29)
(30)
Same definition as in Table 3
1 if department 𝑖 is at location 𝑙; 0 otherwise
Same definition as in Table 3
The integer variable of department 𝑖 at location 𝑗 in
arrangement 𝑘 of location 𝑙
Same definition as in Table 3
Same definition as in Table 3
Monetary term of locating department 𝑖 at location 𝑗
Same definition as in Table 3
Assumptions
The number of departments is equal to the number of locations
The available space can be divided into equal blocks
Equal sized departments
𝑓𝑖𝑘 is independent of the locations of the departments
𝑐𝑗𝑙 is independent of the departments and that it is cheaper to transport patients directly
from department 𝑖 to department 𝑘 than through a third location
Inputs
Monetary term of locating department 𝑖 at location 𝑗
The integer variable of department 𝑖 at location 𝑗 in arrangement 𝑘 of location 𝑙
Total number of departments/locations
Fixed cost of locating and operating department 𝑖 at location 𝑗
Flow of patients between department 𝑖 and department 𝑘
Cost of transporting one patient between location 𝑗 and location 𝑙
Outputs
Minimised total cost
Assignment of 𝑛 departments to 𝑛 locations
A simple integer programming problem formulation of the QAP was developed by Love and
Wong (1976). This is shown in Table 8. Take note of the assumptions of this formulation. In this
formulation, the positions of departments are defined by rectangular coordinates. It was found that
this model cannot be optimally solved for problems with nine or more departments (Bozer &
Meller, 1997).
LIPPs are more easily solvable than convex non-linear integer programming problems. The difficulty
with solving linear and non-linear integer programming problems is due to the fact that it requires one
to search for a set of feasible integer points in order to solve the problem optimally. Integer
programming problems have many local solutions which means that one would have to prove that a
particular solution is superior to all the other feasible solutions in order to find the global optimum
(Love & Wong, 1976). Linear programming uses a feasible region which forms a convex set. The
convexity of this problem means that any local solution is a global one. For this reason, this study will
further focus on the LIPP formulation in Table 7 rather than nonlinear programming.
28
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
TABLE 8: LINEAR INTEGER PROGRAMMING PROBLEM (LOVE & WONG, 1976)
Objectives
Notation
Minimise travel distance between departments
Assign 𝑛 departments to 𝑛 locations
𝑛
𝑛
𝑛−1
𝑛
𝑟
𝑙
𝑎
𝑏
∑ ∑ 𝑎𝑖𝑗 𝑥𝑖𝑗 + ∑ ∑ 𝑓𝑖𝑘 (ℎ𝑖𝑘
+ ℎ𝑖𝑘
+ 𝑣𝑖𝑘
+ 𝑣𝑖𝑘
)
Minimise
𝑖=1 𝑗=1
Subject to
(31)
𝑖=1 𝑘=𝑖+1
𝑟
𝑙
ℎ𝑖𝑘
− ℎ𝑖𝑘
= 𝑥̅𝑖 − ̅̅̅,
𝑥𝑘 𝑖 = 1,2, … , 𝑛; 𝑘 = 𝑖 + 1,
(32)
𝑎
𝑏
𝑣𝑖𝑘
− 𝑣𝑖𝑘
= 𝑦̅𝑖 − ̅̅̅,
𝑦𝑘 𝑖 = 1,2, … , 𝑛; 𝑘 = 𝑖 + 1,
(33)
𝑛
𝑥̅𝑖 + 𝑦̅𝑖 = ∑(𝑥̅𝑗 + 𝑦̅𝑗 )𝑥𝑖𝑗 , 𝑖 = 1,2, … , 𝑛,
(34)
𝑗=1
𝑛
𝑥̅𝑖 − 𝑦̅𝑖 = ∑(𝑥̅𝑗 − 𝑦̅𝑗 )𝑥𝑖𝑗 , 𝑖 = 1,2, … , 𝑛,
(35)
𝑗=1
𝑛
∑ 𝑥𝑖𝑗 = 1, 𝑖 = 1,2, … , 𝑛,
(36)
𝑗=1
𝑛
∑ 𝑥𝑖𝑗 = 1, 𝑗 = 1,2, … , 𝑛,
(37)
𝑗=1
𝑙
ℎ𝑖𝑘
=
𝑙
ℎ𝑖𝑘
=
𝑎
𝑣𝑖𝑘
=
𝑏
𝑣𝑖𝑘
=
(𝑥̅𝑖 , 𝑦̅𝑙 )
𝑥𝑖𝑗
𝑓𝑖𝑘
=
=
=
𝑥𝑖𝑗 ∈ {0,1}, 𝑖, 𝑗 = 1,2, … , 𝑛,
(38)
𝑟
𝑙
𝑎
𝑏
ℎ𝑖𝑘
, ℎ𝑖𝑘
, 𝑣𝑖𝑘
, 𝑣𝑖𝑘
≥ 0, 𝑖 − 1,2, … , 𝑛; 𝑘 = 𝑖 + 1,
(39)
𝑥̅𝑖 , 𝑦̅𝑖 ≥ 0, 𝑖 = 1,2, … , 𝑛.
(40)
Horizontal distance between department 𝑖 and k when department 𝑖
is to the left of department 𝑘; 0 otherwise
Horizontal distance between departments 𝑖 and k when department
𝑖 is to the right of department 𝑘; 0 otherwise
Vertical distance between departments 𝑖 and k when department 𝑖
is above department 𝑘; 0 otherwise
Vertical distance between departments 𝑖 and 𝑘 when department 𝑖
is below department 𝑘; 0 otherwise
Location of department 𝑖
Same definition as in Table 3
Same definition as in Table 3
Assumptions
The locations are given as points on a two dimensional plane
The transportation costs are proportional to weighted rectangular distances
The locations are specified by rectangular coordinates
Inputs
Horizontal distance between departments 𝑖 and 𝑘 when department 𝑖 is to the right of
department 𝑘
Horizontal distance between departments 𝑖 and 𝑘 when department 𝑖 is to the left of
department 𝑘
Vertical distance between departments 𝑖 and 𝑘 when department 𝑖 is above department 𝑘
Vertical distance between departments 𝑖 and 𝑘 when department 𝑖 is below department 𝑘
Flow of material between department 𝑖 at location 𝑘
Output
Minimised traveling distance between departments
29
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.4.5 MIXED INTEGER PROGRAMMING PROBLEM
A Mixed Integer Programming Problem (MIPP) is a model wherein some of the decision variables are
constrained to be integer values at the optimal solution (Frontline Systems Incorporated, 2015). In
other words, it is an optimisation method that combines continuous and discrete variables. This allows
MIPPs to model complex planning and control problems involving both continuous and discrete
variables (Richards & How, 2005).
According to Heragu and Kusiak (1987), amongst all integer-programming formulations of the QAP,
the MIPP have the smallest number of variables and constraints. The formulation of the MIPP is shown
in Table 9. This formulation has 𝑛2 continuous variables, 𝑛2 zero-one variables, and 𝑛2 + 2𝑛
constraints. The MIPP formulation was proved by Burkard (1984) to be equivalent to the QAP
formulation.
TABLE 9: MIXED INTEGER PROGRAMMING PROBLEM (HERAGU & KUSIAK, 1987)
Objectives
Notation
Minimise total cost
Assign 𝑛 departments to locations
𝑛
Minimise
𝑛
∑ ∑ 𝑤𝑖𝑗
(41)
𝑖=1 𝑗=1
𝑛
Subject to
∑ 𝑥𝑖𝑗 = 1, 𝑖 = 1,2, … , 𝑛,
(42)
𝑗=1
𝑛
∑ 𝑥𝑖𝑗 = 1, 𝑗 = 1,2, … , 𝑛,
𝑖=1
𝑛
(43)
𝑛
𝑒𝑖𝑗 𝑥𝑖𝑗 + ∑ ∑ 𝑏𝑖𝑗𝑘𝑙 𝑥𝑘𝑙 − 𝑤𝑖𝑗 ≤ 𝑒𝑖𝑗 ,
𝑘=1 𝑙=1
𝑛
(44)
𝑛
𝑤𝑖𝑗 = 𝑥𝑖𝑗 ∑ ∑ 𝑏𝑖𝑗𝑘𝑙 𝑥𝑘𝑙 ,
(45)
𝑘=1 𝑙=1
𝑛
𝑛
𝑒𝑖𝑗 = ∑ ∑ 𝑏𝑗𝑘𝑙 ,
(46)
𝑘=1 𝑙=1
𝑥𝑖𝑗
𝑥𝑘𝑙
𝑐𝑗𝑙
𝑓𝑖𝑘
𝑡𝑖𝑗
𝑏𝑖𝑗𝑘𝑙
𝑛
=
=
=
=
=
=
=
𝑓𝑖𝑘 𝑐𝑗𝑙 + 𝑡𝑖𝑗
𝑖 = 𝑘, 𝑗 = 1
𝑏𝑖𝑗𝑘𝑙 = {
𝑓𝑖𝑘 𝑐𝑗𝑙
𝑖 ≠ 𝑘 𝑜𝑟 𝑗 ≠ 𝑙
(47)
𝑤𝑖𝑗 ≥ 0, 𝑖 = 1,2, … , 𝑛,
(48)
𝑥𝑖𝑗 ∈ {0,1}, 𝑖, 𝑗 = 1,2, … , 𝑛.
(49)
Same definition as in Table 3
Same definition as in Table 3
Same definition as in Table 3
Same definition as in Table 3
Same definition as in Table 3
Same definition as in Table 3
Same definition as in Table 3
30
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
MIXED INTEGER PROGRAMMING PROBLEM (HERAGU & KUSIAK, 1987) (CONTINUED)
Assumptions
𝑓𝑖𝑘 is independent of the locations of the departments
𝑐𝑗𝑙 is independent of the departments and that it is cheaper to transport patients directly from
department 𝑖 to department 𝑘 than through a third location
𝑡𝑖𝑗 includes gross revenues minus cost of primary input but does not include the
transportation cost of patients between departments
Inputs
Monetary term of locating department 𝑖 at location 𝑗
Flow of material from department 𝑖 to department 𝑘
Total number of departments
Fixed cost of locating and operating department 𝑖 at location 𝑗
Flow of patients between department 𝑖 and department 𝑘
Cost of transporting one patient between location 𝑗 and location 𝑙
Outputs
Minimised total cost
Assign 𝑛 departments to locations
A large MIPP was formulated by Ritzman, Bradford and Jacobs (1979) for finding the optimal
arrangement of offices in a building. By developing a computer program they were able to evaluate the
layout solutions generated against six objectives (Shouman, Nawara, Reyad & El-Darandaly, 2001).
A few years later Montreuil (1991) developed another MIPP formulation that defines binary decision
variables for north-south and east-west relationships. The constraints of this formulation prevent that
departments overlap, have a specific size and shape. A disadvantage of Montreuil’s model (as with all
the other FLP formulations) is the very limited problem size that can be solved optimally.
Shouman et al. (2001) found that this model is limited to six departments. Meller et al. (2001)
tightened Montreuil’s model using valid inequalities and as a result managed to solve problems with
up to eight departments. According to Konak, Kulturel-Konak, Norman and Smith (2006) Montreuil’s
model is a pioneering formulation for the MIP. Montreuil’s model is not based on the traditional QAP
formulation like the one in Table 9. The Linear Mixed Integer Model (LMIM) discussed later on in this
section (Section 2.4.8) is a specialised case of Montreuil’s model (Meller & Gau, 1996). The difference
between these two models is that the LMIM formulation requires that the departments’ lengths and
widths need to be known a priori which improves the computation time when solving this model.
Because these two models are very similar, only the LMIM is further discussed in this study.
Konak, Kulturel-Konak, Norman and Smith (2006) developed a MIPP formulation that is based on the
Flexible Bay Structure (FBS). This formulation allows departments to be placed only in parallel. Each
bay’s width can vary which depends on the total area of the departments in that specific bay. The
departments are restricted to only one bay and are bounded by straight aisles on either side.
Consequently, the FBS restricts possible layout configurations and the complexity of the problem is
reduced (Chang & Lin, 2012). An advantage of using this formulation is that optimal FBS layouts were
found with up to fourteen departments. However, these restrictions imposed by FBS formulations may
be useful in manufacturing cells or the arrangement of rooms, but large departments of a hospital are
31
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
unlikely to be optimally arranged in parallel bays. In order to explore more layout configurations, this
formulation is not further considered in this study.
2.4.6 GRAPH THEORETIC PROBLEM
The Graph Theoretic Problem (GTP) constructs a graph that maximises the weight on the adjacencies
between pairs of departments. It was first formulated by Fould and Robinson (1976) who defined each
department as a node inside a graph network. Initially, areas and shape of departments are ignored.
Knowledge about the possible locations for the departments is not required in advance which makes
the GTP suitable for solving layouts with unequal areas (Kim & Kim, 1995). This graph network relies
on the desirable adjacency of each pair of departments. It is thus assumed that the desirability of
locating each pair of departments adjacent to each other is known (Singh & Sharma, 2006). In this
context, a graph is formed by connecting a set of nodes by arcs, termed vertices and edges respectively,
and represented as 𝐺 = (𝑉, 𝐸) whereby 𝑉 pertains to the set of vertices and 𝐸 pertains to the set of
edges. The areas bounded by cycles of edges are referred to as the faces of the graph. The region
outside the graph is also a face, but is considered an infinite face. Any two faces that have a common
edge are defined as adjacent faces. A graph is weighted if its edges are assigned weights, that is,
parameters such as cost or benefit. A graph is defined as planar when all its edges only intersect at the
vertices. Furthermore, if a graph is fully connected, i.e. contains the maximum number of edges
without losing planarity, then it is a Maximal Planar Graph (Hassan & Hogg, 1987). This formulation is
shown in Table 10.
TABLE 10: GRAPH THEORETIC PROBLEM (HERAGU & KUSIAK, 1987)
Objectives
Notation
Maximise a closeness rating measure
Assign 𝑛 departments to locations
∑ ∑ 𝑟𝑖𝑗 𝑜𝑖𝑗
Maximise
(50)
𝑖∈𝐸 𝑗∈𝐸
Subject to
𝐺 = (𝑉, 𝐸)
=
𝑟𝑖𝑗
=
𝑉
𝐸
𝑁
=
=
=
𝐹
=
𝑜𝑖𝑗
=
𝐸′
=
𝑜𝑖𝑗 = 1, {𝑖, 𝑗} ∈ 𝑁,
(51)
𝑜𝑖𝑗 = 0, {𝑖, 𝑗} ∈ 𝐹.
(52)
A weighted graph with 𝑉 as a nonempty set of departments, 𝐸 as a
set of edges disjoint from 𝑉
The closeness rating indicating desirability of locating department 𝑖
adjacent to department 𝑗
Nonempty set of departments
A set of edges disjoint from V
The set of pairs of departments which must be adjacent in any
feasible solution
The set of pairs of departments which must not be adjacent in any
feasible solution
Variable indicating assignment of departments. 1 if department 𝑖 is
adjacent to department 𝑗; 0 otherwise
{𝑖, 𝑗}: 𝑥𝑖𝑗 = 1, [𝑖, 𝑗] ∈ 𝐸
32
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
GRAPH THEORETIC PROBLEM (HERAGU & KUSIAK, 1987) (CONTINUED)
Assumptions
The desirability of locating each pair of departments adjacent to each other is known
Inputs
The closeness rating indicating desirability of locating department 𝑖 adjacent to department 𝑗
The set of pairs of departments which must be adjacent in any feasible solution
The set of pairs of departments which must not be adjacent in any feasible solution
The set of departments
Outputs
Maximised closeness
Adjacency of departments
The Graph Theoretic approach usually follows three steps, namely (Kim & Kim, 1995; Meller &
Gau, 1996):

Developing a Maximal Planar Weighted Graph (MPWG) (Maximal Planar Graph with the maximum sum
of edge weights) from department relationships to indicate which departments will be adjacent;

Constructing the dual graph of the MPWG to represent departments as adjacent regions having specific
boundaries; and

Converting the dual graph into a block layout where departments have regular shapes with specific
areas.
These steps of developing the MPWG and its dual correspond to the selection and placement steps of
other existing facilities layout procedures (Kim & Kim, 1995). It is essentially a construction approach
(modelling the FLP), but improved solutions have also been obtained using this approach.
The GTP has several advantages over other existing FLP approaches. Firstly, there is the question of
the establishment of an upper bound on the optimal solution. Since the maximum number of
adjacencies that can be achieved in a planar graph is 3𝑛 − 6, summing the highest weight 3𝑛 − 6
relationships provides the required bound (Heragu & Kusiak, 1987). Secondly, the value of the
objective function obtained as the sum of weights of edges is in general better than those obtained by
traditional computerised procedures such as ALDEP and CORELAP. Both of these computerised
procedures achieve adjacencies by constructing planar weighted graphs from a spanning tree or
maximum spanning tree while Graph Theory depends on constructing a MPWG and thus more
relationships are likely to be satisfied resulting in an enhanced objective function value (Hassan &
Hogg, 1987). Thirdly, Foulds and Giffen (1985) showed that it is possible to impose an alternative
objective during the solving process.
Despite these advantages of the GTP, there are several limitations too. For one, the actual length of
common boundaries of adjacent departments is not taken into consideration when constructing the
block plan or calculating the objective function via the MPWG (Hassan & Hogg, 1987). While the
approach has succeeded in finding good arrangements of departments, it has failed to find good
configurations for them (arrangements in a particular form) (Kim & Kim, 1995).
33
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.4.7 LINEAR CONTINUOUS MODEL
The Linear Continuous Model (LCM) was developed by Heragy and Kusiak (1991). They formulated
single-row as well as multi-row problems for both equal areas and unequal areas. Single-row and
multi-row layout problems are also known as one-dimensional and two-dimensional space allocation
problems respectively. Since multi-row is a better representation of real life problems only the multirow layout problem will be discussed. Heragy and Kusiak (1991) found that the algorithms can be
solved in a relatively low computation time with good quality sub-optimal solutions. According to their
research the LCM is more useful in solving FLPs than other models promoted in literature.
v-axis
Department
i
hi
Department
j
hj
vi
vj
h-axis
FIGURE 8: VARIABLES FOR EQUAL-SIZED LINEAR CONTINUOUS MODEL
Figure 8 shows the variables of the Equal-sized Linear Continuous Model formulation. Table 11 shows
the formulation for a FLP with equal areas.
In order to transform the formulation to solve unequal areas, the following variables are defined in
addition to those already defined in Table 11:

𝑙𝑖 = the length of the horizontal side of department 𝑖; and

𝑏𝑖 = the length of the vertical side of department 𝑖.
The variables for the unequal-sized LCM are described in Figure 9.
li
v-axis
gi
Department
i
lj
hi
gj
hj
Department
j
vi
vj
h-axis
FIGURE 9: VARIABLES FOR UNEQUAL-SIZED LINEAR CONTINUOUS MODEL
34
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
TABLE 11: LINEAR CONTINUOUS MODEL FOR EQUAL-SIZED DEPARTMENTS (HERAGY & KUSIAK, 1991)
Objectives
Notation
Minimise the total cost
Assign 𝑛 departments to locations
𝑛−1
Minimise
𝑛
∑ ∑ 𝑐𝑖𝑗 𝛼𝑖𝑗 (|ℎ𝑖 − ℎ𝑗 | + |𝑣𝑖 − 𝑣𝑗 |)
(53)
𝑖=1 𝑗=𝑖+1
|ℎ𝑖 − ℎ𝑗 | + |𝑣𝑖 − 𝑣𝑗 | ≥ 1,
𝑖 − 1, … , 𝑛 − 1; 𝑗 = 1, … , 𝑛; 𝑥𝑖 , 𝑦𝑖 𝑖𝑛𝑡𝑒𝑔𝑒𝑟; 𝑖 = 1, … , 𝑛,
|ℎ𝑖 − ℎ𝑗 | ≤ ℎ − 1, 𝑖 = 1, … , 𝑛 − 1; 𝑗 = 𝑖 + 1, … , 𝑛,
Subject to
𝑛
ℎ𝑖
ℎ𝑗
𝑣𝑖
𝑣𝑗
𝑐𝑖𝑗
𝛼𝑖𝑗
𝑣
ℎ
=
=
=
=
=
=
=
=
=
(54)
(55)
|𝑣𝑖 − 𝑣𝑗 | ≤ 𝑣 − 1, 𝑖 = 1, … , 𝑛 − 1; 𝑗 = 𝑖 + 1, … , 𝑛.
(56)
Same definition as in Table 3
The horizontal distance between department 𝑖 and the v-axis
The horizontal distance between department 𝑗 and the v-axis
The vertical distance between department 𝑖 and h-axis
The vertical distance between department 𝑗 and h-axis
Cost of transporting one patient between location 𝑗 and location 𝑙
The frequency of trips required between department 𝑖 and 𝑗
Vertical dimensions of the floor plan
Horizontal dimensions of the floor plan
Assumptions
The departments are arranged on a two dimensional plane as shown in Figure 8
The departments are oriented in only two given directions
The shape of the departments is known in advance
There is no restriction on the shape of the building in which the departments are to be
located
Inputs
The horizontal distance between department 𝑖 and the v-axis
The horizontal distance between department 𝑗 and the v-axis
The vertical distance between department 𝑖 and h-axis
The vertical distance between department 𝑗 and h-axis
The cost per unit distance travelled between departments 𝑖 and 𝑗
The cost per unit distance travelled between departments 𝑖 and 𝑗
The number of trip to be made between departments 𝑖 and 𝑗
Vertical dimensions of the floor plan
Horizontal dimensions of the floor plan
Outputs
Minimised total cost
Assign 𝑛 departments to locations
Heragy and Kusiak (1991) formulated another version of the LCM where constraints are added to
ensure that departments do not overlap. The formulation of this model is shown in Table 12.
The difference between the model in Table 11 and the one in Table 12 is that the last model may lead
to a layout solution with empty spaces between departments (Solimanpur & Jafari, 2008). The model
is thus suitable for solving specific types of layout problems. The variable 𝑀, an arbitrarily large
positive number, used in (58) and (59) ensures that no two departments in the layout overlap.
35
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
TABLE 12: LINEAR CONTINUOUS MODEL FOR UNEQUAL SIZED DEPARTMENTS (HERAGY & KUSIAK, 1991)
Objectives
Notation
Minimise transportation cost (material handling cost)
Assign 𝑛 departments to locations
𝑛−1
Minimise
𝑛
∑ ∑ 𝑐𝑖𝑗 𝛼𝑖𝑗 (|𝑥𝑖 − 𝑥𝑗 | + |𝑦𝑖 − 𝑦𝑗 |)
(57)
𝑖=1 𝑗=𝑖+1
Subject to
1
(𝑙 + 𝑙𝑗 ) + 𝑘𝑖𝑗 , 𝑖 = 1, … , 𝑛 − 1;
2 𝑖
𝑗 = 𝑖 + 1, … , 𝑛,
1
|𝑦𝑖 − 𝑦𝑗 | + 𝑀(1 − 𝑧𝑖𝑗 ) ≥ (𝑔𝑖 + 𝑔𝑗 ) + 𝑘𝑖𝑗 , 𝑖 = 1, … , 𝑛 − 1;
2
𝑗 = 𝑖 + 1, … , 𝑛,
|𝑥𝑖 − 𝑥𝑗 | + 𝑀𝑧𝑖𝑗 ≥
𝑧𝑖𝑗 (1 − 𝑧𝑖𝑗 ) = 0, 𝑖 = 1, … , 𝑛 − 1; 𝑗 = 𝑖 + 1, … , 𝑛.
𝑛
ℎ𝑖
ℎ𝑗
𝑣𝑖
𝑣𝑗
𝑐𝑖𝑗
𝛼𝑖𝑗
𝑧𝑖𝑗
=
=
=
=
=
=
=
=
𝑀
𝑙𝑖
𝑙𝑗
𝑔𝑖
𝑔𝑗
𝑘𝑖𝑗
=
=
=
=
=
=
(58)
(59)
(60)
Same definition as in Table 3
The horizontal distance between department 𝑖 and the v-axis
The horizontal distance between department 𝑗 and the v-axis
The vertical distance between department 𝑖 and h-axis
The vertical distance between department 𝑗 and h-axis
Cost of transporting one patient between location 𝑗 and location 𝑙
The frequency of trips required between department 𝑖 and 𝑗
Variable used to ensure that only one of the constraints hold so that
the departments do not overlap
An arbitrarily large positive number
The distance between the extreme vertical sides of department 𝑖
The distance between the extreme vertical sides of department 𝑗
The distance between the extreme horizontal sides of department 𝑖
The distance between the extreme horizontal sides of department 𝑗
Minimum distance by which departments 𝑖 and 𝑗 are to be separated
Assumptions
The departments are arranged on a two dimensional plane as shown in Figure 8
The departments are oriented in only two given directions
The shape of the departments is known in advance
The departments are square or rectangular in shape.
There is no restriction on the shape of the building in which the departments are to be
located
Inputs
Total number of departments
Cost of transporting one patient between location 𝑗 and location 𝑙
The number of trips to be made between department 𝑖 and 𝑗
The distance between the extreme vertical sides of department 𝑖
The distance between the extreme vertical sides of department 𝑗
The distance between the extreme horizontal sides of department 𝑖
The distance between the extreme horizontal sides of department 𝑗
Minimum distance by which departments 𝑖 and 𝑗 are to be separated
Outputs
The horizontal distance between department 𝑖 and the v-axis
The horizontal distance between department 𝑗 and the v-axis
The vertical distance between department 𝑖 and h-axis
The vertical distance between department 𝑗 and h-axis
Minimised total cost
Assign 𝑛 departments to locations
36
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.4.8 LINEAR MIXED INTEGER MODEL
According to Heragy and Kusiak (1991) the Linear Mixed Integer Model (LMIM) has a lower number of
integer variables than any other formulation of the FLP (with the exception of the LCM). Most LMIMs
available in the literature have been obtained through linearizing the QAP, however the LMIM
discussed in this section is not. The location of sites does not need to be known prior to solving the
problem. Heragy and Kusiak (1991) also found that this formulation delivers high quality solutions in
a relatively short time. This model takes time into consideration while the LCM focused on cost.
The variables for the unequal-sized LMIM problem are described in Figure 10.
sijh
li
v-axis
gi
Department
i
lj
sijv
hi
gj
Department
j
vi
vj
hj
h-axis
FIGURE 10: VARIABLES FOR UNEQUAL-SIZED LINEAR MIXED INTEGER MODEL
Table 13 shows the formulation for the LMIM:
TABLE 13: LINEAR MIXED INTEGER MODEL (HERAGY & KUSIAK, 1991)
Objectives
Notation
Minimise the total time involved in patients moving between departments
Assign 𝑛 departments to locations
𝑛−1
𝑛
∑ ∑ 𝑡𝑖𝑗 𝛼𝑖𝑗 (|𝑥𝑖𝑣 − 𝑥𝑗𝑣 | + |𝑥𝑖ℎ − 𝑥𝑗ℎ |)
Minimise
(61)
𝑖=1 𝑗=𝑖+1
1
(𝑏 + 𝑏𝑗 ) + 𝑐𝑖𝑗𝑣 , 𝑖 = 1, … , 𝑛 − 1;
2 𝑖
𝑗 = 𝑖 + 1, … , 𝑛,
1
|𝑥𝑖ℎ − 𝑥𝑗ℎ | + 𝑀(1 − 𝑦𝑖𝑗 ) ≥ (𝑙𝑖 + 𝑙𝑗 ) + 𝑐𝑖𝑗ℎ ,
2
𝑖 = 1, … , 𝑛 − 1; 𝑗 = 𝑖 + 1, … , 𝑛,
|𝑥𝑖𝑣 − 𝑥𝑗𝑣 | + 𝑀𝑦𝑖𝑗 ≥
Subject to
𝑧𝑖𝑗 (1 − 𝑧𝑖𝑗 ) = 0, 𝑖 = 1, … , 𝑛 − 1; 𝑗 = 𝑖 + 1, … , 𝑛.
𝑡𝑖𝑗
=
𝛼𝑖𝑗
=
𝑠𝑖𝑗𝑣
=
(62)
(63)
(64)
The time required for a patient to move between departments 𝑖
and 𝑗
Same definition as in Table 12
Vertical clearance, i.e. the minimum distance by which
departments 𝑖 and 𝑗 are to be separated if they are positioned in
opposite rows in the layout
37
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
LINEAR MIXED INTEGER MODEL (HERAGY & KUSIAK, 1991) (CONTINUED)
Notation
𝑠𝑖𝑗ℎ
=
𝑙𝑖
𝑙𝑗
𝑏𝑖
𝑏𝑗
ℎ𝑖
ℎ𝑗
𝑣𝑖
𝑣𝑗
𝑧𝑖𝑗
=
=
=
=
=
=
=
=
=
The horizontal clearance, i.e. the minimum distance by which
departments 𝑖 and 𝑗 are to be separated if they are positioned in
the same rows in the layout
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Same definition as in Table 12
Assumptions
The departments are arranged on a two dimensional plane
The departments are oriented in only two given directions
The shape of the departments is known in advance
There is no restriction on the shape of the building in which the departments are to be
located
Inputs
The time required by a material handling carrier per trip between departments 𝑖 and 𝑗
The frequency of trips required between departments 𝑖 and 𝑗
The vertical clearance
The horizontal clearance
The distance between the extreme horizontal sides of department 𝑖
The distance between the extreme vertical sides of department 𝑖
Vertical distance between centre of department 𝑖 and h-axis
Horizontal distance between centre of department 𝑖 and v-axis
Outputs
Minimised total time
Departments assigned to locations
2.5 Solution methods
As mentioned in the introduction to FLP solution methods in Section 2.3.2, the layout models
discussed in Section 2.4 can be particularly time-consuming to solve for real-word (i.e. large)
problems. They can be optimally solved with exact methods such as the Branch and Bound Algorithm
and the Cutting Plane Algorithm, but only if the problem size is small enough since the computation
time significantly increases as the problem size increases. Typically a hospital layout will be too large
to be solved with exact methods (Tompkins, White, Bozer & Tanchoco, 2010). Consequently, heuristics
and metaheuristics have been developed so as to search for near optimal solutions of large FLPs. These
methods are able to find good quality solutions within reasonable computation times which are
sufficient for practical purposes. The Centre for Operational Research and Logistics (2015) of
Portsmouth University in England found heuristic and metaheuristic techniques to be powerful and
flexible search methodologies that have successfully tackled practical and difficult problems. Many
heuristics have been developed to specifically solve large FLPs, such as: CRAFT, ALDEP, CORELAP,
SPIRAL, and MULTIPLE. They are popular as layout software and are discussed in Section 2.6. The field
38
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
of metaheuristics has enjoyed considerable popularity in the last few decades and these methods are
currently being used to solve large FLPs (Blum, Roli & Sampels, 2008). Examples of metaheuristics
used to solve FLPs are shown in Table 1 and the more popular metaheuristics were chosen to be
analysed, namely: SA, TS, GA, ACO, and PSO. A description, history, algorithm outline, flow diagram,
and the advantages and disadvantages of these methods are provided in this section. As mentioned in
Section 2.3, hybrid metaheuristics, a new branch of metaheuristic research concerned with the
hybridisation of metaheuristics with algorithmic components originating from other techniques, has
also started to gain research attention. These methods are briefly discussed in Section 2.5.3. The
chapter ends with a conclusion on the analysed methods.
2.5.1 EXACT METHODS
Exact methods are able to find the optimal solution of problems. There exist two categories of exact
methods, namely Branch and Bound Algorithms and Cutting Plane Algorithms.
Branch and Bound Algorithms
Branch and Bound Algorithms can be utilised to find an optimal solution to the QAP, since the QAP has
only binary variables. Literature reports optimal solutions only up to a problem size of 16 (Singh &
Sharma, 2006). Beyond this problem size it becomes intractable for even a powerful computer to
handle. The first Branch and Bound Algorithms were proposed separately by Gilmore (1962) and
Lawler (1963). These two methods differ with regards to the computation of the lower bounds. Each of
the methods implicitly evaluate all possible solutions. Land (1953) and Gavett and Plyter (1966)
developed two more algorithms which involves the assignment of pairs of departments whereas the
first algorithms assign one department to one location. In comparison to the original algorithms which
assign one department to one location, these ones assign pairs of departments to pairs of
locations (Charnsethikul, 1988).
All four algorithms operate by assigning departments to locations at each phase of the solving process.
Each phase involves backtracking, excluding some assignments and resuming of the search process.
Lavalle and Roucairol (1985) suggest that parallel Branch and Bound Algorithms can be used to solve
the QAP optimally. The algorithms search in a parallel fashion through numerous parts of the decision
tree. However, the authors found that high computation time is required for layouts with 12 or more
departments (Drira, Pierreval & Hajri-Gabouj, 2007). Burkard proposed in 1973 that the QAP can be
solved via reducing its square matrix. This implies that the matrix is changed into a matrix with nonnegative elements, with at least one zero being present in every row and column. Furthermore, this
reduction increases the influence of the linear term in the objective function and the quality of the
bound via decreasing the magnitude of the quadratic term in the function. Bazaraa (1975) proposed
another algorithm which suggests a partial layout at each phase of the solving process. Furthermore, it
determines a lower bound on the cost on all potential layouts that can arrive from these partial
39
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
layouts. If it is determined that the lower bound has a lower cost than the best found layout, the
algorithm will continue to assign new departments. Otherwise, the search process along this path will
be terminated and a new search path is then followed. The search process will continue until a whole
layout is arrived at. Bazaraa and Elshafei (1979) developed a Branch and Bound Algorithm that
assigns only one department to locations at each phase of the search process. Kaku and
Thompson (1986) suggested another Branch and Bound Algorithm which performed better than the
first developed algorithm (Heragu & Kusiak, 1987).
The problem with exact methods is that the optimal solution may be found early in the search process,
but it is not accepted until a large number of solutions are computed (Burkard & Stratman, 1978;
Bazaraa & Kirca, 1983). Exact methods thus require high memory as well as high computation time.
Therefore in 1984, researchers developed a heuristic which terminates the search process early
without verifying that the optimal solution is arrived at. The criteria for the termination process
include time limits (stopping the search of a specified period of time), and quality of the upper bounds
(after a predefined time limit the upper bound is decreased if there is no improvement in the
solution) (Burkard, 1984).
Cutting Plane Algorithms
Cutting Plane Algorithms refine an objective function by means of linear inequalities (called cuts)
(Mitchell, 1998). The Cutting Plane Algorithm gained popularity in 1980 when Bazaraa and
Sherali (1980) established an algorithm that is built on Benders’ partitioning scheme (which allows
one to compute the solution of very large linear programming problems that have a special block
structure) (Benders, 1962). Three years later Burkard and Bonninger (1983) established another
Cutting Plane Algorithm to solve the QAP. The optimal Cutting Plane Algorithm has a high computation
time and its storage is complex. Heragu and Kusiak (1987) found that the largest size FLP that can be
optimally solved by a Cutting Plane Algorithm is one with only eight departments. These Cutting Plane
Algorithms are known as traditional cutting planes. Another type is the so-called polyhedral cutting
planes which are used in complex Branch and Cut Algorithms (involves combining the Branch and
Bound and Cutting Plane Algorithm). According to Cela (2013) the experience of applying this method
to the QAP is still very limited. This is due to the structural combinatorial properties of the QAP.
2.5.2 METAHEURISTICS
Metaheuristics
have
recently
become
very
popular
for
solving
optimisation
problems
(Bianchi, Dorigo, Gambardella & Gutjahr, 2009). They combine heuristics in a more general
framework. However, they do not ensure that a globally optimum solution can be found. Many
metaheuristics use some form of stochastic optimisation, so that the solution is dependent on a set of
random variables generated.
40
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
According to Blum and Roli (2003), metaheuristics are:
…high level concepts for exploring search spaces by using different strategies. These strategies
should be chosen in such a way that a dynamic balance is given between the exploitation of the
accumulated search experience (which is commonly called intensification) and the exploration of
the search space (which is commonly called diversification).
Finding a balance between the ‘intensification’ and ‘diversification’ is necessary to quickly determine
areas in the search space that exhibit high quality solutions while not wasting too much time in search
space areas that are either already explored or provide low quality solutions (Blum & Roli, 2003).
The field of metaheuristics originated in 1983 with the development of SA which as the name implies,
simulates the annealing process of metals (Kirkpatrick, Gelatt & Vecchi, 1983). Next, the Artificial
Immune System was created by Farmer, Packard and Perelson (1986). After this Laguna, Barnes and
Glover (1991) developed TS and pioneered the use of memory in metaheuristics whereby the search
moves are recorded and avoided in further search steps. Ant Colony Optimisation (ACO) was
introduced by Dorigo (1992) who was inspired by the swarm intelligence of social ants using
pheromones as a chemical messenger. Koza (1992) published in the same year research on genetic
programming. PSO was then developed through Kennedy and Eberhart (1995). The next year Storn
and Price (1996) constructed the Differential Evolution, a vector based evolutionary algorithm. Geem,
Kim and Loganathan (2001) came up with the Harmony Search Algorithm which was inspired by
music. After this, Passino (2002) developed the Bacteria Foraging Algorithm. Nakrani and
Tovey (2004) optimised internet hosting centres when they invented the Honey Bee Algorithm. This
was further built upon by Karaboga (2005) with his design of the Artificial Bee Colony Algorithm, and
later his Firefly Algorithm. Additionally, Yang and Deb (2014) developed the Cuckoo Search Algorithm.
Metaheuristics can be divided into single-solution based and population based searches (Talbi, 2009).
Single-solution based metaheuristics are also called S-metaheuristics, as the name suggests that they
seek to improve a single solution while solving optimisation problems. They can be seen as ‘walks’
through neighbourhoods or search trajectories through the solution space of the problem. These
trajectories or walks are conducted through iterative search procedures that move from the current
solution to other ones in the solution space (Blum & Li, 2008). With each iteration, generation and
replacement procedures are applied to the current single solution. The generation procedures involve
determining a set of candidate solutions from a current solution, ‘s’. The set of candidate solutions,
C(s), are generated by local transformations of the current solution. The replacement procedures on
the other hand involve selecting solutions from the candidate solutions in order to replace the current
solution. The process will continue until a specified stopping criterion is reached. The generation and
replacement phases may be memoryless. If this is the case, the generation and replacement are based
only on the current population. Otherwise, history of the search stored in memory can be used in both
41
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
the generation and replacement phases. Popular examples of S-metaheuristics include LS, SA, and TS.
All three algorithms will be discussed in this section. On the other hand, population based
metaheuristics, also called P-metaheuristics, can be viewed as an iterative improvement in a
population of solutions (candidates) (Kennedy, Kennedy, Eberhart & Shi, 2001). The candidate
solutions are improved by using population characteristics. The first step is to initialise the population.
Next, a new population is generated in what is known as the generation phase. Finally, this new
population is integrated into the current one through selection procedures called the replacement
phase. The search process terminates when a given condition is satisfied, i.e. the stopping criterion.
Similar to S-metaheuristics, the generation and the replacement phases can be memoryless. Examples
of this class of metaheuristics include ACO, Artificial Immune Systems, Bee Colony, Evolutionary
Algorithms, Estimation of Distribution Algorithms, Particle Swarm Optimisation (PSO), and Scatter
Search.
There exists another category of metaheuristics which uses the behaviour of decentralised, selforganising optimisation called Swarm Intelligence. Swarm Intelligence can be described as algorithms
inspired from the collective behaviour of species such as ants, bees, birds, fish, termites, and
wasps (Blum & Mercle, 2008). It originated form the social behaviour of those species that compete for
foods. (Talbi, 2009) states that “the main characteristics of swarm intelligence-based algorithms are
particles are simple and non-sophisticated agents, they cooperate by an indirect communication medium,
and do movements in the decision space.” Two very popular and successful swarm intelligence
approaches include ACO and PSO.
Metaheuristics can be divided into two types of search strategies, one that is an improvement on
simple LS algorithms, or one that has a learning component to the search. The improvement types
include SA, TS, LS, variable neighbourhood search and GRASP. The learning component type on the
other hand includes metaheuristics, such as ACO, evolutionary computation, and GAs (Blum &
Roli, 2003).
For the purposes of this study it is sufficient to look into the following: the LS, SA, TS, GA, ACO, and
PSO.
Local Search method
LS is one of the first popular improvement techniques developed and also one of the most basic
iterative heuristic methods, which typically produces suboptimal solutions (Lourenço, Martin &
Stützle, 2010). It is able to approximate solutions to large scale combinatorial optimisation problems
in a short period of time. As it always moves from the current solution to the next better solution, it is
called a greedy algorithm. When an algorithm is greedy it means that it follows the problem solving
heuristic which entails making the locally optimal choice for each and every stage in the hopes of
finding a global optimum as rapidly as possible.
42
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
The algorithm starts at an initial solution from which it iteratively tries to replace the current solution
by moving to an improved solution in a specified neighbourhood of the current solution. This
neighbourhood can be described as follows (Blum, Roli & Sampels, 2008):
…a neighbourhood structure is a function 𝑁: 𝑆 → 2𝑆 that assigns to every 𝑠 ∈ 𝑆 a set of neighbours
𝑁(𝑠)  𝑆. 𝑁(𝑠) is called the neighbourhood of s. Often, neighbourhood structures are implicitly
defined by specifying the changes that must be applied to a solution s in order to generate all its
neighbours. The application of such an operator that produces a neighbour 𝑠′ ∈ 𝑁(𝑠) of a solution s
is commonly called a move.
The topology of the search landscape is defined by the problem instance. One can view the search
landscape as a labelled graph whereby the nodes represent solutions and the arcs represent the
neighbourhood relations that exist between solutions. A solution 𝑠 ∗∈ 𝑆 is called a global minimum if
for all 𝑠 ∈ 𝑆 it holds that 𝑓(𝑠 ∗) ≤ 𝑓(𝑠). The set of all global minimum is thus denoted by 𝑆 ∗. The
concept of local minimum can be defined as follows (Blum et al., 2008) “a local minimum with respect to a
neighbourhood structure 𝑁 is a solution 𝑠̂ such that ∀ 𝑠 ∈ 𝑁(𝑠̂ ): 𝑓(𝑠̂ ) ≤ 𝑓(𝑠). We call
ŝ a strict locally minimal
solution if 𝑓(𝑠̂ ) < 𝑓(𝑠) ∀ 𝑠 (𝑠̂ ).”
The iterative improvement search is the most basic search method and is shown below in
Algorithm 2.1. Each consecutive move is performed only if the resulting solution is superior to the
current solution. Once it has reached a local minimum, the algorithm terminates.
ALGORITHM 2.1: ITERATIVE IMPROVEMENT LOCAL SEARCH ALGORITHM ADAPTED FROM BLUM, ROLI AND SAMPELS (2008)
Iterative improvement local search
1 𝑠 ← Generate initial solution
2 while ∃ 𝑠′ ∈ 𝑁(𝑠) such that 𝑓(𝑠′) < 𝑓(𝑠) do
3
𝑠 ← Choose improving neighbour from 𝑁(𝑠)
4 end while
There are also a few metaheuristics that are based on the LS such as Variable Neighbourhood Search,
Guided LS, and Greedy Randomized Adaptive Search Procedure (Blum et al., 2008).
Simulated Annealing
Simulated Annealing (SA) exhibits the ability to escape local optima, which significantly improves the
prospects for finding the global optimum solution. The goal of SA is thus to find an acceptably good
solution in a set amount of time (Dowsland & Thompson, 2012). SA is a popular direct search
technique most commonly used to address discrete problems (as well as continuous optimisation
problems, but to a lesser extent) (Pham & Karaboga, 2012). SA is applicable to unconstrained and
bound-constrained optimisation problems with the ability to manage several variables. Direct search
methods have a tendency to converge slower, but are also more tolerant to the presence of noise in the
objective function and constraints. SA is a good solver for problems with general non-specific problem
structures (Dowsland & Thompson, 2012).
43
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
The SA algorithm was inspired by the annealing process in metal work which involves heating and
cooling a metal so as to alter its physical properties through changes in the internal molecular
structure (Aarts, Korst & Michiels, 2014). Once cooled, the metal structure becomes fixed and this
lower energy internal structure is then retained. If the cooling and annealing process is too quick, it
will solidify into a sub-optimal configuration. Instead, if the metal cools slowly, the crystals within the
metal will solidify optimally into a state of minimum energy. The temperature is the variable which
simulates the heating and cooling process. Typically the temperature starts off high and slowly
decreases due to cooling as the algorithm progresses. When the temperature is high, the algorithm
allows more frequently for the accepted solutions to be worse off than the current solution so as to
enable the algorithm to escape any local optima it finds itself in at the start of the execution. Escaping
the local optima is facilitated by the algorithm allowing hill-climbing moves which lead to worse
objective function values with the aim of finding the global optimum. As the temperature drops, so
does the probability of worse solutions being accepted and thus the algorithm gradually focuses in on
a search space area where there is a high chance of a close to optimum solution existing (Suman &
Kumar, 2006). Due to gradual ‘cooling’, the algorithm is remarkably effective when solving large
problems with numerous local optimums. The state of certain physical systems, as well as the
objective function, are equivalent to the internal energy of the said system in the specified state. The
goal of SA can therefore be seen as bringing the system from an arbitrary initial state to a state with
the minimum possible level of energy (Aarts et al., 2014).
The following terminology will be used to describe the SA algorithm:





𝑥𝑖
𝐸
𝑇
∆
𝑖
Design vector, i.e. design, architecture, configuration;
Objective function value, i.e. system energy;
Variable, i.e. system temperature;
Difference in system energy between two design vectors; and
Iteration number (𝑖 = 1,2,3 …).
The Simulated Annealing Algorithm is as follows:
1.
2.
3.
4.
5.
Choose a random 𝒙𝒊 , select the initial system temperature, and specify the cooling (i.e. annealing)
schedule: similar to the LS method, SA utilises multiple starting points and finds an optimum starting
point from each of them. In order to improve optimisation, a temperature must be selected that will
initially allow for almost any move away from the current solution to be acceptable. This allows the
algorithm to better explore the whole search space prior to cooling whereby it will settle into a more
confined region;
Evaluate 𝑬(𝒙𝒊 ) using a simulation model: the initial value of the objective function is found in order to
provide a basis for comparison with neighbouring locations in the proceeding iterations;
Perturb 𝒙𝒊 to obtain a neighbouring design vector 𝒙𝒊+𝟏 : for each iteration a new point is generated
randomly. The distance between the new point and the current point (the extent of the search) is based
on a probability distribution with a scale proportional to the temperature;
Evaluate 𝑬(𝒙𝒊+𝟏 ) using a simulation model: a new value of the objective function is found in order to
determine if it is better or not;
If 𝑬(𝒙𝒊+𝟏 ) < 𝐸(𝒙𝒊 ) , then 𝒙𝒊+𝟏 is the new current solution: the algorithm accepts all new points that
lower the value of the objective function;
44
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
6.
If (𝒙𝒊+𝟏 ) > 𝐸(𝒙𝒊 ) , then accept 𝒙𝒊+𝟏 as the new current solution with an acceptance probability of
−∆𝑬
𝒆 𝑻 where ∆= 𝑬(𝒙𝒊+𝟏 ) − 𝑬(𝒙𝒊 ): points that increase the value of the objective function are accepted
with a certain probability. Such points allow the algorithm to avoid being trapped in local minima which
enables the algorithm to search globally for more possible solutions. The acceptance probability
−∆𝑬
7.
8.
function 𝒆 𝑻 determines which solutions to accept. Once the neighbour solution has been found to be a
worse solution, a couple of factors are regarded, namely: by how much is the neighbour solution worse,
and; how high is the current temperature of the system. At a high temperature the system is more likely
to accept solutions that are worse than the current one. Basically, the smaller the change in energy (the
quality of the solution), and the higher the temperature, the higher are the changes for the algorithm to
accept a worse solution;
Reduce the system temperature according to the cooling schedule: the annealing schedule
systematically reduces the temperature as the algorithm proceeds. As the temperature drops, the
algorithm reduces the extent of its search to converge to a minimum; and
Terminate the algorithm: once an optimal solution which satisfies the predefined criteria, or
conversely when the maximum number of iterations is reached, the algorithm terminates and the
optimal or near-optimal solution is arrived at.
The algorithm outline for SA is shown in Algorithm 2.2, and the strengths and weaknesses of SA are
shown in Table 14.
TABLE 14: STRENGTHS AND WEAKNESSES OF SIMULATED ANNEALING ADAPTED FROM DOWSLAND AND THOMPSON (2012) AND VECCHI (1983)
Strengths
Able to deal with highly non-linear models, chaotic and noisy data and many constraints
Is a robust and a general technique
Its flexible nature and ability to approach global optimality are its main benefits over other LS methods
The algorithm is quite versatile since it does not rely on any restrictive properties of the model
Easily adapted to different problem contexts. The ability to fine-tune a given algorithm for use in more than one
problem is very useful in industry
A ‘weak’ method which does not make use of gradient information and makes few assumptions
It can deal with highly non-linear problems and non-differentiable functions as well as functions with multiple
local optima
Excellent at avoiding the problem of getting stuck in local optima and is much better on average at finding an
approximate global optimum, whereas a hill climber algorithm will only accept neighbour solutions that are
better than the current solution
Usually better than greedy algorithms, when it comes to problems that have numerous locally optimum solutions
Statistically guaranteed to converge in asymptotic time to an optimal solution.
Relatively easy to code, even for complex problems
Amenable to parallel implementation
Can deal with arbitrary systems and cost functions
Weaknesses
Since it is a metaheuristic, a lot of decisions are required to turn it into an actual algorithm
It is computation intensive
Tailoring work is required to account for different classes of constraints and to fine-tune the parameters of the
algorithm can be rather delicate
Clear trade-off between the quality of the solutions and the time is required to compute them
Precision of the numbers used in implementation of SA can have a significant effect upon the quality of the
outcome
Repeatedly annealing with a schedule is very slow, especially if the cost function is difficult to compute (which
also probably makes it expensive)
Heuristic methods, which are problem-specific or take advantage of extra information about the system, will
often be better than general metaheuristic methods, although SA is often comparable to heuristics
Computation time required in SA grows exponentially with respect to the size of the problem and this may end
up in SA requiring more iterations than Exhaustive Search methods
Cannot tell whether it has found an optimal solution. Some other complimentary method (e.g. Branch and Bound
Algorithm) is required to do this
45
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
ALGORITHM 2.2: SIMULATED ANNEALING ALGORITHM OUTLINE ADAPTED FROM BERTSEKAS (1999)
Simulated annealing algorithm outline
1 Generate initial feasible solution
2 while System not solidified (i.e. T > 0) do
3 |
while Thermodynamic equilibrium not reached do
4 |
| Generate neighbouring solution (i.e. perturb current solution)
5 |
| Evaluate the change in energy ∆𝐸 resulting from the perturbation
6 |
| if ∆𝐸 < 0 then
7 |
| | Accept the move and make the neighbouring solution the current solution
8 |
| else
9 |
| | Accept the move with probability 𝑒 −∆𝐸
𝑇
10
11
12
13
14
|
| end
|
end
|
Lower temperature according to the cooling schedule
end
Report incumbent solution
Tabu Search
Tabu Search (TS) incorporates non-improving moves in an improving search algorithm in the hopes of
finding the global optimum. TS was proposed by Glover (1986) who then formalised it in 1989. The
main idea behind TS is to act similar to LS when it gets close to a local optimum. The basic argument
behind this algorithm is to pursue local search by allowing non-improving moves whenever the
algorithm encounters a local optimum. LS methods have a tendency to become stuck in suboptimal
regions. Cycling back to formerly visited solutions (i.e. being stuck in local minima) is disallowed by
making use of a memory list, appropriately called the tabu list, which keeps a record of the recent
search history (Gendreau, 2003). After each iteration of the algorithm, a group of moves that includes
those which would return immediately to the previous point are added to the tabu list. No such move
is allowed for a few iterations, but eventually all moves are removed from the tabu list and are made
available once again.
The following definitions will be used to describe the TS algorithm:






𝑥
𝑥̂
𝑖
𝑖𝑚𝑎𝑥
𝑇
𝑁(𝑥)
Set of variables (solution set);
Incumbent solution;
Iteration number (𝑖 = 1,2,3, …);
Maximum iterations;
Tabu list; and
Neighbourhood set of a solution 𝑥.
The Tabu Search Algorithm is as follows:
1.
2.
Choose a random feasible solution 𝒙(𝒐) , and an iteration limit 𝒊𝒎𝒂𝒙 , and set the incumbent
̂ ← 𝒙(𝒐) as well as the solution index 𝒊 ← 𝟎: initialise the search by choosing a feasible
solution 𝒙
solution and deciding on the number of iterations. No moves are listed as tabu yet. The search will now
enter a loop in the hopes of finding a superior feasible solution;
Stop the search process if no non-tabu move leads to a feasible neighbour of current solution or if
the iteration limit is reached: stop the search if no non-tabu move ∆𝑥 in move set 𝑁(𝑥) leads to a
46
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
3.
4.
5.
6.
7.
feasible neighbour of current solution 𝑥 (𝑖) , or if 𝑖 = 𝑖𝑚𝑎𝑥 . Take note that incumbent solution 𝒙
̂ is an
approximate optimum;
Choose a non-tabu feasible move: select any non-tabu feasible move ∆𝑥 ∈ 𝑁 as ∆𝑥 (𝑖+1) ;
Update the solution set: update 𝑥 (𝑖+1) ← 𝑥 (𝑖) + ∆𝑥;
Update the incumbent solution if it is necessary: if the objective function value of 𝑥 (𝑖+1) is superior to
that of incumbent solution 𝑥̂ , replace 𝑥̂ with 𝑥 (𝑖+1) ;
Update the tabu list: add a collection of moves that includes any returning immediately from 𝑥 (𝑖+1) to
𝑥 (𝑖) . Remove from the tabu list any forbidden moves that have been on it for a sufficient number of
iterations; and
Increment the iteration number: increment 𝑖 ← 𝑖 + 1 and return to step 2.
ALGORITHM 2.3: TABU SEARCH ALGORITHM OUTLINE ADAPTED FROM GENDREAU (2003)
Tabu Search algorithm outline
1 Generate initial feasible solution
2 Set tabu list 𝑇 ← ∅
3 while Stopping criterion not met do
Generate current solution neighbourhood 𝑁(𝑥)
4
Evaluate the performance of non-tabu solutions in 𝑁(𝑥)
5
Select best neighbour solution 𝑥̂
6
7
Update incumbent solution (if necessary)
Update tabu list 𝑇
8
Set new current solution 𝑥 ← 𝑥̂
9
10 end
11 Report incumbent solution
The algorithm outline for TS is shown in Algorithm 2.3, and the strengths and weaknesses of TS are
shown in Table 15.
TABLE 15: STRENGTHS AND WEAKNESSES OF TABU SEARCH ADAPTED FROM HANAFI (2001) AND ABIDO (2002)
Strengths
Can be directly applied to virtually any kind of optimisation problem
Uses a flexible memory of search history to prevent cycling and to avoid entrapment in local optima
Can easily deal with non-convex, non-smooth, and non-differentiable objective functions
Has been theoretically proven that TS algorithm can yield global optimal solution with a probability of one
Able to avoid entrapment in local optimal solution and prevent cycling by using flexible memory of search history
Uses payoff (performance index or objective function) information to guide the search in the problem space.
Therefore, it can easily deal with non-smooth, non-continuous, and non-differentiable objective functions that are
the real-life optimisation problems. Additionally, this property relieves TS of assumptions and approximations,
which often are required by traditional optimisation methods for many practical optimisation problems
Uses probabilistic transition rules to make decisions, not deterministic rules. Hence, TS is a kind of stochastic
optimisation algorithm that can search a complicated and uncertain area to find the global optimum. This makes
TS more flexible and robust than conventional methods
Robustness to the initial solution. Therefore, TS can be used to improve the solution quality obtained by other
classical techniques
Allows non-improving solution to be accepted in order to escape from a local optimum
Can be applied to both discrete and continuous solution spaces
For larger and more difficult problems (scheduling, quadratic assignment, and vehicle routing), TS obtains
solutions that rival and often surpass the best solutions previously found by other approaches
Weaknesses
Too many parameters to be determined
Number of iterations could be very large
Global optimum may not be found, depends on parameter settings
47
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
Genetic Algorithms
Genetic algorithms (GA) became popular through the work of Holland (1973) especially after
publishing a book on GAs in 1975 (Rajeev & Krishnamoorthy, 1992). The algorithm combines Darwin’s
principle of survival of the fittest with structured information exchange using randomised operators to
evolve an efficient search mechanism. GAs efficiently exploit useful information contained in a
population of solutions to generate new solutions with an improved performance. Goldberg and
Samtani (1986) applied the GA to optimise the design of a 10-bar truss and Deb (1991) applied GA to
optimising the design of welded joints.
GAs are a related class of algorithms to the improvement procedures in that it begins with a set of
possible solutions and attempts to improve it, but with using analogies to natural processes. It uses
mutation and crossover techniques to evolve the existing solutions into better ones. Being based on
the principal of survival of the fittest, it uses a selection scheme biased towards fitter individuals to
make up the next generation. After a few generations the algorithm will converge with the best
individual which represents hopefully the optimum. GAs have the following features:





An initial population of solutions;
A mechanism for generating new solutions by combining features from solutions in the existing
population;
A mechanism for generating a new solution by operating on a single previously known solution;
A mechanism for selecting the set of solutions from the populations, giving preference to those with
better objective function values; and
A mechanism for removing solutions from the population.
The solutions generated during iterations can be selected to mutate or to reproduce. The selection
process is biased towards choosing better solutions in the current population. The mutation that
occurs for the FLP takes the form of some variant of the pair-wise exchange. Any activity assigned to
the same location will occupy that location in the offspring. The activity assignments of the remaining
locations are chosen randomly from one or the other parent. Next, the unassigned activities are
matched with the remaining unassigned locations. As new offspring are created, the solutions with the
poorest values of the objective are eliminated in order to keep the population size the same.
The following definitions will be used to describe the Genetic Algorithm:







𝑝
𝑥
𝑝𝑒
𝑝𝑖
𝑝𝑐
𝑖
𝑖𝑚𝑎𝑥
Population size;
Feasible solution set;
Population elites;
Population immigrants;
Population crossovers;
Generation index (𝑖 = 1,2,3 …); and
Maximum generation index.
The Genetic Algorithm is as follows:
1.
Choose any feasible solution, population size, generation limit, and population subdivisions:
initialise the search process by choosing population size 𝑝, a set of feasible solutions 𝑥 (1) , …, 𝑥 (𝑝) ,
48
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.
3.
4.
5.
6.
generation limit 𝑖𝑚𝑎𝑥 , population size 𝑝𝑒 , population immigrants 𝑝𝑖 , and population crossovers 𝑝𝑐 . Also
set generation index 𝑖 ← 0;
Stop the search process if the specified generation limit is reached: stop the search process if
𝑖 = 𝑖𝑚𝑎𝑥 and report the best solution of the current population as an approximate optimum;
Choose elite individuals of the population: initialise the new generation 𝑖 + 1 with copies of the 𝑝𝑒
best solutions in the current generation. By doing this, the best genes are passed on to the next
generation. The purpose of this is to improve the gene pool and thus the objective function value;
Arbitrarily choose new immigrant feasible solutions: randomly select 𝑝𝑖 new immigrant feasible
solutions and include them in the 𝑖 + 1 population. This step allows new genes to enter into the
population in the hopes of finding a superior solution;
Execute crossovers and add them to the new generation: choose 𝑝𝑐 /2 non-overlapping pairs of
solutions from the generation 𝑖 population, and execute crossover on each pair at an independently
chosen random cut point to complete the generation 𝑖 + 1 population; and
Increment the generation index: increment 𝑖 ← 𝑖 + 1 and return to step 2.
ALGORITHM 2.4: GENETIC ALGORITHM OUTLINE ADAPTED FROM RAJEEV AND KRISHNAMOORTHY (1992)
Genetic algorithm outline
1 Generate initial population of individuals
2 Evaluate the fitness of each individual
3 while Stopping criterion not met do
4
Select parent solutions for reproduction
5
Generate offspring population
6
Apply mutation operator to offspring solutions
7
Evaluate fitness of offspring solutions
8
Select individuals to be carried over to the next generation
9
Update incumbent solution (if necessary)
10 end
11 Report incumbent solution
The algorithm outline for GA is shown in Algorithm 2.4, and the strengths and weaknesses of GA are
shown in Table 16.
TABLE 16: STRENGTHS AND WEAKNESSES OF GENETIC ALGORITHMS ADAPTED FROM RAJEEV AND KRISHNAMOORTHY (1992) AND SIMPSON, DANDY AND
MURPHY (1994)
Strengths
Not limited by restrictive assumptions about search space, such as continuity or existence of derivatives
Does not require problem-specific knowledge to carry out a search. For instance, calculus-based search
algorithms use derivative information to carry out a search. In contrast to this, GAs are indifferent to problemspecific information
Works on coded design variables, which are finite length strings. These strings represent artificial chromosomes.
Every character in the string is an artificial gene. GAs process successive populations of these artificial
chromosomes in successive generations
Uses a population of points at a time in contrast to the single-point approach by the traditional optimisation
methods. That means, at a given time, GAs process a number of designs
Uses randomized operators in place of the usual deterministic ones. The random operators improve the search
process in an adaptive manner
Deals directly with a population of solutions at any one time. These are spread throughout the solution space, so
the chance of reaching the global optimum is increased significantly
Each solution consists of a set of discrete sizes. One does not have to round diameters up or down to obtain the
final solution
Identifies a set of solutions of network configurations that are close to the minimum cost solution. These
configurations may correspond to quite different designs that can be then compared in terms of other important
but non-quantifiable objectives
49
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
STRENGTHS AND WEAKNESSES OF GENETIC ALGORITHMS ADAPTED FROM RAJEEV AND KRISHNAMOORTHY (1992) AND SIMPSON, DANDY AND MURPHY
(1994) (CONTINUED)
Uses objective function or fitness information only, compared with the more traditional methods that rely on
existence and continuity of derivatives or other auxiliary information
Can solve every optimisation problem which can be described with the chromosome encoding
Solves problems with multiple solutions
Since the GA execution technique is not dependent on the error surface, one can solve multi-dimensional, nondifferential, non-continuous, and even non-parametrical problems
Structural GA gives us the possibility to solve the solution structure and solution parameter problems at the same
time by means of GA
Very easy to understand and it does not demand the knowledge of advanced mathematics
Easily transferred to existing simulations and models
Weaknesses
Certain optimisation problems (they are called variant problems) cannot be solved by means of GAs. This occurs
due to poorly known fitness functions which generate bad chromosome blocks in spite of the fact that only good
chromosome blocks cross-over
There is no absolute assurance that a GA will find a global optimum. It happens very often when the populations
have a lot of subjects
Like other artificial intelligence techniques, the GA cannot assure constant optimisation response times. Even
more, the difference between the shortest and the longest optimisation response time is much larger than with
conventional gradient methods. This unfortunate GA property limits the GA’s use in real time applications
GA applications in controls which are performed in real time are limited because of random solutions and
convergence, in other words this means that the entire population is improving, but this could not be said for an
individual within this population. Therefore, it is unreasonable to use GAs for on-line controls in real systems
without testing them first on a simulation model
Ant Colony Optimisation
The first Ant Colony Optimisation (ACO) algorithm proposed was the Ant System (AS) by Dorigo in his
doctoral dissertation (Dorigo, Maniezzo & Colorni, 1996). AS has been successfully applied to
relatively small instances of the Travelling Salesman Problem (TSP), up to and including 75 cities. It
exhibits an equal capacity to evolutionary approaches to compute the results for general-purpose
heuristics. However, AS did not prove able in competing with state-of-the-art algorithms that were
explicitly designed for large instances of the TSP. AS originally consisted of a set of three algorithms,
namely ant-cycle, ant-density, and ant-quantity. For ant-density and ant quantity, ants update the
pheromone directly after each move from a city to an adjacent city. In contrast to this, ant-cycle
updates the pheromone only once all ants have constructed the trails, with the amount of pheromone
deposited by each ant being set to be a function of the tour (path) quality. Since ant-cycle produced
superior results over the other two variants, it was called Ant System and the other two algorithms
were no longer studied. Dorigo’s proposal stimulated a number of researchers to make extensions and
improvements to its basic ideas. Dorigo and Di Caro worked together to develop the commonly
accepted definition of the ACO (Di Caro & Dorigo, 1998). ACO has been applied to combinatorial
problems and has proven effective in solving various problems, such as scheduling, routing, and
assignment (Dorigo & Stützle, 2003).
The main principle behind ACO algorithms is to mimic the cooperative behaviour of real ants in order
to solve optimisation problems (Dorigo, Di Caro & Gambardella, 1999). These algorithms can be seen
as multi-agent systems wherein each and every agent is inspired by the behaviour of a real ant. The
50
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
reason why a real ant’s behaviour is of interest is that ants use collective behaviour to perform
complex tasks, including food transportation and determining shortest paths to food sources. These
ACO algorithms are built on the principle that a communication mechanism which is very simple can
lead to finding the shortest path between points.
Figure 11 illustrates an experiment with a real colony of ants. Take note that the ants cannot see the
obstacle very well. The colony moves along a path between the food source and the colony nest and
during their trips they communicate indirectly with one another via ground trails of odorous volatile
substances known as pheromones. The role of the trails is to guide the other ants toward the target
point, i.e. food or nest. The larger the amount of pheromone on a certain path, the larger the
probability that the ants will follow that path. According to Talbi (2009), “for a given ant, the path is
chosen according to the smelt quantity of pheromone.” Furthermore, pheromone evaporates over time
and the quantity left by one ant depends on the amount of food (reinforcement process). As Figure 11
illustrates, when an ant reaches an obstacle on its path, there exists an equal probability for each and
every ant to choose to go left or to go right around it. Since the left trail is shorter than the right trail, it
requires less travel time and thus the ant will end up leaving a higher level of pheromone. The more
ants that take the left path, the higher the level of pheromone, and hence the shortest path emerges
(Rardin, 1998).
Nest
Nest
Nest
Nest
Food
Food
Food
Food
FIGURE 11: BEHAVIOUR OF ANTS SEARCHING FOR AN OPTIMAL PATH BETWEEN THE FOOD AND THEIR NEST ADAPTED FROM DORIGO, MANIEZZO AND
COLORNI (1996)
The following definitions will be used to describe the ACO:








𝐸
𝑖
𝑖𝑚𝑎𝑥
𝑙
𝑎
𝑘
𝐺(𝑉, 𝐸)
𝑇𝑎𝑘
Trail intensity factor;
Iteration number (𝑖 = 1,2,3, … 𝑖𝑚𝑎𝑥 );
Maximum iteration number;
Iteration number (𝑙 = 1,2,3 …);
Each ant in a set of ants;
Node;
Ant colony with set 𝑉 of nodes and set 𝐸 of edges inter-linking all pairs of nodes; and
List of remaining nodes to be visited by ant 𝑎 given that it is currently at node 𝑘.
51
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
The ACO algorithm is as follows:
1.
2.
3.
4.
5.
6.
Choose a random set 𝑨 of ants, and an iteration limit 𝒊𝒎𝒂𝒙 , set the iteration number 𝒊 ← 𝟎, and
assign an initial pheromone level to every edge in 𝑮(𝑽, 𝑬): initialise the search process by choosing a
set 𝐴 of ants, deciding on the maximum number of iterations, and assigning pheromone levels to all the
edges. The search process will now enter into an iteration loop in the hopes of generating an optimal
solution;
Stop the search process if the iteration limit is reached: stop the search if 𝑖 = 𝑖𝑚𝑎𝑥 ;
Assign an initial random node 𝒌 ∗∈ 𝑽 to ant 𝒂 and update the list of nodes to be visited 𝑻𝒂𝒌 ←
𝑽\{𝒌 ∗}: assign an initial random node 𝑘 ∗∈ 𝑉 to ant 𝑎 (𝑎 ∈ 𝐴) and update 𝑇𝑎𝑘∗ . Set an iteration number
𝑙 ← 0;
Choose node 𝒋 and update pheromone concentration as well as the set of nodes to be visited:
continue with this step while < |𝑉|. Choose 𝑗 ∈ 𝑇𝑎𝑘 according to the rule of displacement. Update the
pheromone concentration on edge (𝑘 ∗, 𝑗). Update the set of nodes to be visited 𝑇𝑎𝑘 ← 𝑇𝑎𝑘∗ \{𝑗}. Set 𝑘 ∗←
𝑗 and increment the iteration number 𝑙;
Update the incumbent solution if it is necessary: perform global pheromone update on incumbent
solution edges; and
Increment the iteration number: increment 𝑖 ← 𝑖 + 1 and return to step 2.
ALGORITHM 2.5: ANT COLONY OPTIMISATION ALGORITHM OUTLINE ADAPTED FROM DORIGO AND STÜTZLE (2003)
Ant colony optimisation algorithm outline
1 Assign every edge in 𝐺(𝑉, 𝐸) an initial pheromone level
2 𝑖←0
3 Generate a set 𝐴 of ants
4 while 𝑖 < 𝑖𝑚𝑎𝑥 do
for all 𝑎 ∈ 𝐴 do
5
Assign an initial random node 𝑘 ∗∈ 𝑉 to ant 𝑎
6
𝑇𝑎𝑘 ← 𝑉\{𝑘 ∗}
7
𝑙←0
8
while 𝑙 < |𝑉| do
9
Choose a node 𝑗 ∈ 𝑇𝑎𝑘 according to the rule of displacement
10
Update the pheromone concentration on edge (𝑘 ∗, 𝑗)
11
Update the set of nodes to be visited 𝑇𝑎𝑘 ← 𝑇𝑎𝑘∗ \{𝑗}
12
𝑘 ∗← 𝑗
13
𝑙 ← 𝑙+1
14
15
end
16
end
17
Update incumbent solution
18
Perform global pheromone update on incumbent solution edges
𝑖 ←𝑖+1
19
20 end
21 Report incumbent solution
The algorithm outline for ACO is shown in Algorithm 2.5, and the strengths and weaknesses of ACO are
shown in Table 17.
TABLE 17: STRENGTHS AND WEAKNESSES OF ANT COLONY OPTIMISATION ADAPTED FROM BOSWAL, PODDAR AND SHEKHAWAT (2009)
Strengths
Inherent parallelism
Positive Feedback accounts for rapid discovery of good solutions
Efficient for Traveling Salesman Problem and similar problems
Can be used in dynamic applications (e.g. adapts to changes such as new distances)
Fast, solutions of reasonable quality
52
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
STRENGTHS AND WEAKNESSES OF ANT COLONY OPTIMISATION ADAPTED FROM BOSWAL, PODDAR AND SHEKHAWAT (2009) (CONTINUED)
Weaknesses
Theoretical analysis is difficult
Sequences of random decisions (not independent)
Probability distribution changes by iteration
Research is experimental rather than theoretical
Time to convergence uncertain (but convergence is guaranteed)
Solution may be far from optimum
Generate only limited number of different solutions
Decisions made at early stages reduce a set of possible steps at latter stages
Particle Swarm Optimisation
The implicit rules that members of fish schools and bird flocks follow that result in them being able to
move in a synchronised manner without colliding, and thus resulting in a perfect choreography, was
studied by Reynolds in 1987 and Heppner and Grenander in 1990 (Mandal, Mukhopadhyay & Pal,
2016). Reynolds described the process in three simple behaviours, namely separation, alignment, and
cohesion. According to Wilson (1975), the social behaviour of animals is governed by simple rules. In
contrast to this, human social behaviour is more complex than a flock’s movement. The core idea
behind developing Particle Swarm Optimisation (PSO) is that numerous examples coming from nature
enforce the view that social information sharing between individuals of a population may result in an
evolutionary advantage (Eberhart et al., 1996).
PSO is classified as a P-metaheuristics inspired from swarm intelligence. It mimics the social behaviour
of natural organisms, e.g. birds flying in a flock migrating and fish swimming in a school searching for
food (Talbi, 2009). In such swarms, individual behaviour is coordinated with the group using local
movements without there being any form of central control. A particle swarm can be seen as a cellular
automata, whereby individual cell (particles in PSO) updates are done in parallel. The reason for this is
that each new cell value depends only on the old value of the cell and its neighbourhood, and all cells
are updated using the same rules (Rardin, 1998). PSO combines self-experiences with social
experiences. This means that each particle in the search space adjusts its ‘flying’ according to its own
flying experience as well as the flying experiences of other particles. The particles move towards a
promising area to get the global optimum. Each particle keeps track of its personal best solution
(𝑝𝑏𝑒𝑠𝑡 ) and the best value of any particle/global best (𝑔𝑏𝑒𝑠𝑡 ). Thus, each particle modifies its position
according to its 𝑝𝑏𝑒𝑠𝑡 , 𝑔𝑏𝑒𝑠𝑡 , current position and current velocity. Figure 12 shows a swarm of
particles which move to the global optimum after a certain amount of PSO iterations. After three
iterations it is clear that the particles are moving towards the local and global optimums which are
shown by the darker colours. After seven iterations, all the particles reach the global optimum.
53
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
Initial velocities of swarm particles
Particle velocities after 3 iterations
Particle velocities after 5 iterations
Particle velocities after 7 iterations
FIGURE 12: PARTICLE SWARM OPTIMISATION EXAMPLE ADAPTED FROM HEPPNER AND GRENANDER (1990)
A swarm is made up of 𝑁 particles moving around in a D-dimensional search space. Each particle 𝑎𝑖 is
a solution candidate to the problem and is represented by the vector in the decision space. Each
particle has its own unique position and velocity (flying direction and set of the particle). SPO involves
taking advantage of the ability of the particles to cooperate, similar to birds in a flock or fish in a
school. The success of certain particles influences the behaviour of other particles. Each particle’s
position is successively adjusted toward the global optimum by taking into account two factors,
namely the best position visited by itself and the best position visited by swarm as a whole.
The following definitions will be used to describe the Particle Swarm algorithm:

𝑁
The number of particles in the swarm, i.e. number of birds, fish, etc.;

𝑆
𝑝𝑖
Solution space;
Position of particle 𝑎𝑖 in the solution space;
𝑣𝑖
𝑣(𝑎𝑖 )
Velocity of particle’s 𝑎𝑖 ;
Neighbourhood of particle at 𝑎𝑖 (fixed);
𝑐1
𝑐2
𝑝𝑏𝑒𝑠𝑡
Weight of local information;
Weight of global information;
Best position of the particle; and
𝑔𝑏𝑒𝑠𝑡
Best position of the swarm.







The Particle Swarm algorithm is as follows:
1.
2.
3.
4.
Choose an initial feasible swarm population with random locations and velocity vectors: create a
population (swarm) of particles uniformly distributed;
Evaluate each particle’s position according to the objective function: evaluate each particle’s
current position;
If a particle’s current position is better than its previous best position, update it: determine if the
particle’s current position is better than its previous position and move it to a new position;
Determine the best particle (according to the particle’s best previous positions): determine the
particle with the best position;
54
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
5.
6.
7.
Update particle’s velocities: update particle’s velocities with the following equation 𝑣𝑖𝑡+1 = 𝑣𝑖𝑡 +
𝑡
𝑡
𝑐1 𝑈1𝑡 (𝑝𝑏𝑒𝑠𝑡
− 𝑝𝑖𝑡 ) + 𝑐2 𝑈2𝑡 (𝑔𝑏𝑒𝑠𝑡
− 𝑝𝑖𝑡 ). The first term represents the particle’s initial movement, the
𝑖
second represents the particle’s personal influence and the third term social influence of other particles;
Move particles to their new positions: update the particles’ positions using the following equation
𝑝𝑖𝑡+1 = 𝑝𝑖𝑡 + 𝑣𝑖𝑡+1 ; and
Increment the iteration number: increment 𝑖 ← 𝑖 + 1 and return to step 2 until the stopping criteria
are satisfied.
ALGORITHM 2.6: PARTICLE SWARM ALGORITHM OUTLINE ADAPTED FROM RARDIN (1998) AND ROMERO, ZAMUDIO, BALTAZAR, MEZURA, SOTELO AND
CALLAGHAN (2012)
Particle swarm optimisation algorithm outline
1 Generate initial feasible swarm population with random locations and velocity vectors
2 while Stopping criterion not met do
for all 𝑝 ∈ 𝑆 do
3
4
Evaluate the performance of the particle in objective space
5
Update best location of the particle (if necessary)
6
Update best location in the neighbourhood of the particle (if necessary)
7
Update velocity vector of the particle
8
Accelerate the particle toward a new location
9
end
10
Update incumbent solution (if necessary)
11 end
12 Report incumbent solution
The algorithm outline for PSO is shown in Algorithm 2.6, and the strengths and weaknesses of PSO are
shown in Table 18.
TABLE 18: STRENGTHS AND WEAKNESSES OF PARTICLE SWARM OPTIMISATION ADAPTED FROM TALBI (2009) AND REYNOLDS (1987)
Strengths
Insensitive to scaling of design variables
Simple implementation
Easily parallelised for concurrent processing
Derivative free
Very few algorithm parameters
Very efficient global search algorithm
Weaknesses
Tendency to obtain a fast and premature convergence in mid optimum points
Slow convergence in refined search stage (weak LS ability)
2.5.3 HYBRID METAHEURISTICS
The concept of hybrid metaheuristics has been commonly accepted only in recent years, although the
combination of different metaheuristics strategies dates back to the 1980s (Méndez, 2011). There
exists a generalised consensus on the benefits of integrating various components of different search
methods as well as the widespread use of hybrid techniques in fields of operations research and
artificial intelligence (Blum & Roli, 2008). The main motivation behind hybridisation is to exploit the
complementary character of different optimisation strategies. An adequate combination of
complementary algorithmic concepts can be key for achieving top performance in solving hard
optimisation problems. However, literature shows that it is nontrivial to generalise, i.e. a certain
55
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
hybrid might work well for specific problems, but might perform poorly for others (Blum, Puchinger,
Raidl & Roli, 2011). It is also a difficult task to develop an effective hybrid approach that requires
expertise from different areas of optimisation.
There exists two categories of hybrid metaheuristics. The first consists of designing a solver including
components from more than one metaheuristic and the second combines metaheuristics with other
techniques. The second category typically works with fields such as artificial intelligence and
operations research. Examples include the amalgamation of metaheuristics with constraint
programming, data mining techniques, integer programming, and tree-based search methods
(Gendreau & Potvin, 2010). Lim, Yuan and Omatu (2000) found that a hybrid of GA and LS produced
solutions with average objective values within 0.8% of the best known solutions, for several QAP
benchmarks. Tseng and Liang (2006) proposed a hybrid metaheuristic called ANGEL to solve the QAP
which includes the combination of ACO, GA, and a LS method. They tested over a hundred instances of
the QAP and found that ANGEL exhibited a success rate of 90% with respect to finding the optimal
solution.
2.6 Layout software
Various layout software packages have been developed to solve large FLPs. This software uses
computer-based algorithms such as CRAFT, ALDEP, MATCH, CORELAP, SPIRAL, MULTIPLE and
COFA (Singh & Sharma, 2006), otherwise referred to as heuristics (Armour & Buffa, 1963).
Computer-based algorithms can significantly advance the productivity of the layout planner as well as
the value of the final solution through their ability to generate and numerically evaluate numerous
layout alternatives in a very short period of time. Another advantage is that computerised layout
algorithms are also highly effective with regards to performing rapid ‘what if’ analyses. This allows the
user to vary input data or the layout itself, and thus find the most ideal and practical solution.
However, although current computer-based algorithms are useful, they cannot yet replace human
experience and judgement as they have great shortcomings in terms of their ability to address the
qualitative characteristics/components of facility layouts (Tompkins et al., 2010).
Computer-based layout algorithms can be classified as distance based or adjacency algorithms.
MATCH and SPIRAL are examples of adjacency based, while CRAFT, SHAPE, LOGIC, FLEX-BAY and
MULTIPLE are distance based algorithms (Singh & Sharma, 2006). The difference between these two
types of algorithms is seen in their objective functions. Adjacency based algorithms have the objective
function shown in (65).
Maximise
∑ ∑(𝑟𝑖𝑗 ) 𝑥𝑖𝑗
𝑖
𝑗
56
(65)
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
In (65), 𝑥𝑖𝑗 is 1 if department 𝑖 is adjacent to department 𝑗; else 0, and 𝑟𝑖𝑗 is the closeness rating of
department 𝑖 and 𝑗. The philosophy behind (65) is that material handling cost can be significantly
reduced if the two departments in question have adjacent boundaries. Distance based algorithms have
the objective function shown in (66).
𝑛
𝑛
𝑛
𝑛
1
(𝑇𝐶) = ∑ ∑ ∑ ∑ 𝐶𝑖𝑘 𝐷𝑗𝑙 𝑋𝑖𝑗 𝑋𝑘𝑙
2
Minimise
(66)
𝑖=1 𝑗=1 𝑘=1 𝑙=1
𝑖≠𝑘 𝑗≠𝑙
In (66) 𝐶𝑖𝑘 is the material handling cost between department 𝑖 and 𝑘; 𝐷𝑗𝑙 is the distance between
department 𝑗 and 𝑙; 𝑋𝑖𝑗 is 1 if department 𝑖 is assigned to location 𝑗; else 0; and 𝑋𝑘𝑙 is 1 if department 𝑘
is assigned to location 𝑙; else 0. The basic premise behind (66) is that the distance travelled increases
the total traveling cost.
Layout software can also be classified into two main categories namely construction type and
improvement type. Construction-type involves the development of a new layout from scratch whilst
improvement-type generates layout alternatives based on a pre-existing layout (Tompkins et al.,
2010). There are also computerised algorithms that can be used as both construction and
improvement algorithms such as BLOCPLAN and LOGIC. Table 19 shows the classification of a few
computer-based algorithms.
TABLE 19: CLASSIFICATION OF COMPUTER-BASED ALGORITHMS ADAPTED FROM SCHIFFAUEROVA (2014)
Construction algorithms
ALDEP
CORELAP
PLANET
Improvement algorithms
CRAFT
MCRAFT
MULTIPLE
BLOCPLAN
LOGIC
Different computer-based algorithms are shown in Table 1 and Figure 6. A few of the more popular
algorithms were chosen and will be discussed in this section. The algorithms are CRAFT, LOGIC, and
MULTIPLE. In addition to these BLOCPLAN is also included in this discussion because of its unique
properties.
CRAFT
CRAFT (Computerised Relative Allocation of Facilities Technique) was introduced by Armour and
Buffa (1963). It is an example of an improvement-type method and a greedy algorithm (meaning it
repeatedly selects the immediate best choice from a set of alternatives at each step of its execution).
Since it is an improvement algorithm, it is natural that it departs from an initial layout which must be
provided. Firstly, the centroids of the departments are determined for the initial layout. Next, it the
rectilinear distance between pairs of department centroids is calculated and the values are stored in a
57
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
distance matrix. An initial/rough layout cost can then be determined by multiplying each entry in the
from-to matrix with its corresponding entry in the distance matrix and the unit cost matrix. The next
step is to consider all conceivable two-way (or three-way) departmental exchanges and thereafter
identify the best exchange, i.e. the one that produces the largest reduction in the layout costs. After it is
identified, the layout is updated according to the best exchanges possible and then the new
department centroids are computed in conjunction with the new layout cost. This completes the first
iteration. The second iteration begins with CRAFT identifying the most ideal exchange by looking at all
possible two-way (or three-way) exchanges in the layout. This iterative process continues until the
layout costs cannot be further reduced (Tompkins et al., 2010). One of the advantages of using CRAFT
is that it can capture the initial layout with reasonable accuracy. This is rooted in CRAFT’s ability to
accommodate nonrectangular departments or obstacles located anywhere in a possibly
nonrectangular building. However, one of CRAFT’s weaknesses is that it will rarely generate
department shapes that are straight, with uninterrupted aisles. It is also highly path
dependent (Armour, Buffa & Vollman, 1964).
BLOCPLAN
BLOCPLAN was developed by Donaghey and Pire (1990). It arranges the departments in bands and
uses a from-to chart and a relationship chart as input data for representing the flow. The cost
associated with the layout is measured via the adjacency-based or distance-based objective. Each
department in BLOCPLAN occupies exactly one band which are rectangular in shape. The number of
bands are determined by the program and limited to three bands. However, the widths of the bands
can vary. It may be used as a constructive or an improvement-type algorithm. One of the weaknesses
of BLOCPLAN is that it may not be able to capture the initial layout accurately. It is also difficult to
handle non-fixed departments that may have prescribed or fixed shapes.
LOGIC
LOGIC (Layout Optimisation with Guillotine Induced Cuts) was developed by Tam (1992). It can be
used either as a construction or an improvement algorithm. LOGIC is based on dividing the layout into
small portions by executing successive guillotine (straight lines that run from one end of the layout to
the other). These cuts are executed in a series of horizontal or vertical cuts. With each cut, an
appropriate subset of the departments is assigned to either the east-west or north-south side of the
cut. LOGIC constructs a tree to systematically execute and keep tract of the departments. In
comparison to CRAFT, it is generally not straightforward to model fixed departments or obstacles.
However, LOGIC can be applied in non-rectangular buildings provided that the building shape is
reasonable.
58
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
MULTIPLE
MULTIPLE (Multi-floor Plant Layout Evaluation) was developed by Bozer, Meller and
Erlebacher (1994) for the purpose of developing multi-floor layouts. However, it can also be used in
single-floor layouts by setting the number of floors to one and disregarding data requirements
associated with lifts. MULTIPLE is similar to CRAFT in that a from-to chart is used as input data, and
the objective function is identical. The departments are however not limited to rectangular shapes.
Like CRAFT and MULTIPLE it is an improvement algorithm, and as such begins with an initial rough
layout specified by the planner. Improvements are made at each iteration through two-way exchanges
and the exchange that leads to the greatest reduction in layout cost is then selected. The main
difference between CRAFT and MULTIPLE is that MULTIPLE can make exchanges whether the two
departments are adjacent or not. BLOCPLAN can also make any exchange, but it is based on bands, and
thus fixed departments may either shift or change shape. In essence, MULTIPLE maintains CRAFT’s
flexibility while relaxing CRAFT’s constraint on departmental exchanges.
2.7 Analysis of methods
In this chapter various layout models, solution methods, and layout software were discussed. Seven
popular models for FLPs were presented. Unfortunately, the layout models are very hard to solve for
large problem instances. Exact methods, metaheuristics and hybrid metaheuristics are discussed as
means to solve these layout models. The presented layout models and solution methods are now
further analysed.
2.7.1 ANALYSIS OF LAYOUT MODELS
Table 20 shows the seven layout models with their corresponding variables and objectives. The QAP is
frequently used to solve FLPs. However, instances with a problem size larger than 20, within
reasonable limits, are difficult to manage since even a powerful computer by today’s standards cannot
handle these large instances of the problem. Many QAP formulations can only solve FLPs with equalsized departments, which is difficult to justify in practical cases. It may be unrealistic to design a
hospital with equal sized departments since the area of each department will most likely be different
(these areas are determined in Chapter 4). If this is the case, the layout will include unutilised spaces
and consequently incur unnecessary costs and inefficiency. The QAPs that are able to solve FLPs with
unequal-sized departments divide the departments into numerous small grids each with equal areas,
and effectively forbid the separation of said grids of the same department by assigning artificial flow
between them. As a result of the rise in the number of departments with this approach, it makes
solving even small problems with a few unequal-area departments impossible. Bozer et al. (1994) also
proved that this approach is ineffective since it adds a constraint to the shape of the department. The
optimal solution to the QAP formulation when there exist high artificial flows is therefore likely to be
59
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
an inadequately poor FLP solution. Another weakness of the QAP is that if the distance between
locations cannot be determined beforehand, the QAP model cannot be applied.
TABLE 20: ANALYSIS OF LAYOUT MODELS
x
Quadratic Set Covering Problem
x
Minimise total cost
Minimise fixed cost
Maximise a closeness rating measure
Minimise travel time
x
Minimise flow
x
Minimise material handling cost
Fixed cost
x
Quadratic Assignment Problem
Objectives
Travel cost
Travel time
Frequency of trips
Flow between departments
Closeness rating
Layout model
Distance between departments
Variables
x
x
x
x
Linear Integer Programming Problem
x
x
x
x
Mixed Integer Problem
x
x
x
x
Graph Theoretic Problem
x
x
Linear Continuous Model
x
Linear Mixed Integer Model
x
x
x
x
x
In contrast to this, the QSCP is able to solve FLPs with unequal-sized departments which are more
scalable than the QAP. The drawbacks of this method are that the candidate locations for each centre
have to be established beforehand and the specification of the centre shapes might not be practical in
the sense that some centres have L-shapes, but the model assigns rectangular shapes to them.
Therefore, the QSCP appears to be a better choice than the QAP for solving the FLP with unequal-sized
departments.
The LIPP is shown to be equivalent to the QAP in terms of the variables and objective of the model.
Furthermore, it was found that the model cannot be optimally solved for problems with nine or more
facilities. This model also assumes that the costs of transportation are directly proportional to the
weighted rectangular distances.
The MIPP combines continuous and discrete variables. Among the integer-programming formulations
of the QAP, this model has a relatively small number of variables and constraints. Montreuil’s MIPP
model could be solved with eight departments. Konak, Kulturel-Konak, Norman and Smith (2006)
developed a MIPP model which is based on the flexible bay structure and can be solved for FLPs with
up to fourteen departments. However, possible layout configurations are restricted. According to
Thai (2007), the MIPP model does have advantages over the QAP-based models. With the MIPP model,
60
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
departments may have various shapes and sizes. Furthermore, the shape of the departments is
controlled and irregular-shaped departments are not a problem. Irregular-shaped departments can be
defined as departments which are very long, narrow, or non-rectangular.
The GTP relies on having a predefined desirable adjacency for each pair of departments. The areas and
shape are ignored at the beginning. The model assumes that one knows the desirability of locating
each pair of departments adjacent to each other. Firstly, the establishment of an upper bound on the
optimal solution provides an advantage over other existing FLP approaches. Secondly, the value of the
objective function obtained as the sum of weights of edges in the MPWG is generally better than those
obtained by traditional computerised procedures, such as ALDEP and CORELAP. However, a limitation
of this method is the fact that adjacent departments are not taken into consideration when the block
plan is constructed, or when the objective function value is calculated. The approach has succeeded in
finding good department arrangements, but it failed to find good configurations for them.
The LMIM has the least number of variables of any of the other formulations of the FLPs mentioned.
Consequently, a relatively larger problem size (of 30) can be optimally solved which is almost double
the limit of the other discussed layout models. The advantage of this method, as well as the LCM, is that
site locations do not need to be known before it can be solved. However, the department dimensions
are fixed. These two methods can also be used to solve FLPs with unequal areas. Both of these
methods yield good quality sub-optimal solutions in a comparatively short time of computing. The
LCM is similar to the MIPP which is also based on a continuous representation. According
to Thai (2007), the MIPP formulation is more difficult to solve computationally. The LMIM and the
LCM are relatively less studied than the other layout formulations. No improvements or notable
discussions of both of these methods were found to date in literature. Thus, little research has been
conducted on these two models.
In conclusion, the QSCP may be more practical for a hospital layout than the QAP formulations. With
the MIPP, departments may have various shapes and sizes. The advantage of the GTP lies in the fact
that adjacency of each pair of departments are taken into account. The QSCP, LIPP, and MIPP are QAPbased methods, while the LMIM, and LCM are not. The LMIM and LCM yield good quality solutions in a
relatively short time, but require the dimensions of the departments to be known a priori. The layout
models studied in this thesis can be divided into discrete based, and continuous based. The QAP and
QSCP are discrete based while the MIPP, LCM, and LMIM are continuous based. The MIPP and LMIM
formulations were found to be very similar. Both of these models are newer than the rest of the
models and not much research has been conducted on these specific models so far.
In order to select the most appropriate layout model for designing a rural hospital layout, it is
necessary to first fully comprehend the real world problem. For this reason, the layout model is only
selected in Chapter 5 after discussing the context of a rural hospital (in (Chapter 4).
61
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
2.7.2 ANALYSIS OF SOLUTION METHODS
The two main approaches for solving the FLP were found to be exact methods and metaheuristics.
Exact methods can further be sorted into Branch and Bound Algorithms and Cutting Plane Algorithms.
A few types of Branch and Bound algorithms were found, including a parallel Branch and Bound
Algorithm which can solve layouts with 12 or more facilities. It was found that with Branch and Bound
Algorithms the optimal solution is commonly found early on in the branch and bound process. It is not
however verified until a significantly large number of solutions have been computed. All of algorithms
discussed were found to have high memory and computation requirements. Optimal solutions can also
only be determined up to a problem size of 16 at most.
Two main types of Cutting Plane Algorithms were found, namely traditional cutting planes and
polyhedral cutting planes. It was found that the application of the Cutting Plane Algorithms to the FLP
is still very limited. The optimal Cutting Plane Algorithm has a high computation time and its storage is
complex. As a consequence, the largest size FLP that can be optimally solved by a Cutting Plane
Algorithm is one with only eight facilities.
In conclusion, exact methods can be used to solve the layout model optimally and are therefore the
best methods to use, if the problem size is small enough. Some layout models are also a bit more
complex to solve than others. If the FLP cannot be solved using an exact method, the optimal solution
should be approximated using metaheuristics. Six popular metaheuristics were analysed and are
shown in Table 21. Since the most adequate layout model for modelling the FLP for this study is only
selected in Chapter 5, the best suited solution method for solving the FLP can only be selected
afterwards.
TABLE 21: METAHEURISTICS CLASSIFICATION
Metaheuristic
S-metaheuristics
Local search method
x
Simulated Annealing
x
Tabu Search
x
P-metaheuristics
Swarm Intelligence
Genetic Algorithms
x
Ant colony Optimisation
x
x
Particle Swarm Optimisation
x
x
The LS method is greedy and therefore not the best method to use for a problem with multiple local
minima. For this reason, SA and TS are better methods to use to solve the FLP. The P-metaheuristics
require the iterative improvement in a population of solutions, while the S-metaheuristics improve a
single solution which makes the S-metaheuristics easier to use. The P-metaheuristics are not well
suited for fine-tuning structures which are very close to optimal solutions. However, they are quick to
locate the high performance regions of vast and complex search spaces. After a certain amount of time,
62
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 2 | QUANTITATIVE LAYOUT DESIGN METHODS
the population becomes quite uniform and the solutions of the population no longer improve. The
hybrid metaheuristics combine the strengths of both methods, but are not always the best options to
use due to other factors such as time constraints and unnecessary complexity.
Another approach to modelling and solving the FLP is via computer-based algorithms. These
algorithms are adjacency or distance based. BLOCKPLAN and LOGIC are classified as construction and
improvement type algorithms. It was found that LOGIC is generally not as straightforward to model
fixed departments and BLOCKPLAN may not be able to capture the initial layout accurately. CRAFT
and MULTIPLE are both improvement algorithms that use from-to-charts. A disadvantage of CRAFT is
that it is greedy. MULTIPLE was found to be similar to CRAFT with the exception that it can improve
multi-floor layouts. The discussed computer-based algorithms were found to be effective for varying
data input or the layout itself.
2.8 Conclusion
This chapter provides an introduction to operations research and the FLP in particular. Different
quantitative methods to model and solve the FLP are analysed and compared. Computer-based
algorithms are also discussed. The key findings are summarised in the paragraphs that follow.
The QAP, QSCP, LIPP, MIPP, GTP are found to be more popular approaches to modelling the FLP. A less
studied LCM and LMIM formulations are also analysed. Both of these models reported optimal
solutions for relatively larger problem sizes than the rest of the models. Most of these layout models
aim to optimise the material handling cost and include variables such as distance, flow, transportation
cost, and closeness ratings in the equations, with the exception of the GTP which aims to maximise a
closeness rating measure. It was found that the QAP and the LIPP assume equal sized facilities which
may be unrealistic for designing a hospital layout.
The layout models can be optimally solved using exact methods, e.g. Branch and Bound Algorithms,
Cutting Plane Algorithms, only if the problem size is small enough. Otherwise metaheuristics can be
used to approximate the optimal solution.
Computer-based algorithms were found very effective in rapidly performing ‘what if’ analyses based
on varying the input data or the layout itself. However, the currently available computer-based
algorithms generally do not consider the qualitative characteristics of a layout. Some of these
algorithms are adjacency based while others are distance based. These algorithms can further be
classified into construction type, improvement type, or methods that use both.
This concludes the literature study on quantitative research. The next chapter investigates qualitative
methods and hospital design considerations that need to be taken into account when designing the
layout of a hospital.
63
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3
QUALITATIVE LAYOUT DESIGN METHODS AND
HOSPITAL DESIGN CONSIDERATIONS
“All research ultimately has a qualitative grounding.”
Campbell (1994)
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
64
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
3.1 Introduction
In Chapter 2 the quantitative methods for designing the layout of a facility were analysed. The purpose
of this chapter is to analyse qualitative layout design methods found in literature and to determine
what hospital design considerations should be included when designing a hospital layout. In addition,
the relationship between these hospital design considerations and the quantitative methods analysed
in Chapter 2 is determined.
3.2 LAYOUT TYPES
A facility layout should be arranged according to everything needed for production or delivery of
services (Drira, Pierreval & Hajri-Gabouj, 2007). Therefore layouts for a department, machine shop,
manufacturing cell, warehouse, and work centre would logically be different. Designing the layout of a
facility generally depends on the variety of products as well as the production volumes of the
products (Hartl & Preusser, 2009). Thus, if one considers the FLP for a hospital from a production
layout point of view one needs to take the layout requirements of a hospital into account and
understand exactly what it entails. There are four main types of layouts, often defined according to the
general pattern of work flow (Jacobs, Chase & Chase, 2010), namely: process layouts, product layouts,
fixed-position layouts, and hybrid layouts.
Process layouts naturally group activities that are similar in nature together, in departments, typically
according to the function which they perform. As such, they are occasionally called functional layouts.
They are typically in facilities that produce customised low-volume products that may have the need
for a flexible layout, having different sequences of operations and processing requirements when
producing products (Groover, 2007). An example of this can be seen in a machine shop whereby all
drills are typically located in one work centre, drilling machines in their own work centre, and lathes
in yet another work centre. Process layouts are characteristic of job shops, service delivery, and other
intermittent operations that need to serve different customers with different needs. The main
advantage of such a layout is flexibility but this comes at the cost of the layout’s
efficiency (Jacobs et al., 2010). Oftentimes, customers or jobs do not flow through the system in an
orderly manner, movements between departments tend to take too much time, queues often develop,
and backtracking is common. Operations may need to be set up in a unique manner for each and every
customer, in a configuration that is suitable to their particular processing requirements. In a service
facility, process layouts typically require large aisles so that customers may move between
departments freely. The main concern for such a layout pertains to where exactly the departments
should be located in relation to one another. Although each job or customer will follow a different
route through the system, some paths are more common than others.
65
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
Product layouts are based largely on the sequence necessary for processing the part(s) being
produced on the production line. Each and every product must have its own line specially designed to
meet its unique requirements. Flow of work is commonly efficient and orderly, moving consecutively
from one workstation to another all the way down the assembly line until completion of the product at
the end of the line. When the demand is stable and the production volume is high, product layouts are
ideal for activities of mass production that require repetitive operations (Nahmias & Cheng, 2009).
The product or service targets the general market and not a particular customer per se. The main
advantages of product layouts are efficiency and ease of use but this comes at the cost of flexibility on
the production line.
Oftentimes the product is either too fragile or too heavy to move during a project, and this is when
fixed-position layouts become ideal (Brandon-Jones, 2015). Examples of where it is useful include
most aircraft assembly, housing, and shipbuilding projects. With the product remaining stationary for
the entire manufacturing process, equipment, materials, and workers are brought back and forward to
the site where production takes place. The utilisation of equipment is usually low due to the fact that it
is oftentimes less costly to just leave equipment idle at the site where it is used intermittently
(periodically) than to move it back and forth from one location to another. As a result, equipment is
sometimes leased due to its limited use. Workers brought to the work site are typically highly skilled
at performing the specific tasks they are assigned to (Russel & Taylor, 1999).
Hybrid layouts (also called combination layout) modify and/or combine some characteristics of the
three basic layout types just discussed in order to try satisfy the requirements of a particular
situation (Ali, 2015). An example of this can be seen when a firm may have a process layout for the
majority of its process along with an assembly line off to the side. Conversely, a fixed-position layout
may be used for assembling the final product, while the assembly lines are used to produce
components (i.e. subassemblies) that add up to realise the final product (Advameg, 2015),
e.g. manufacturing a commercial aircraft.
A hospital layout is classified as a hybrid layout since it combines process layout and fixed-layout
characteristics (Ali, 2015; Advameg, 2015). Hospitals may have an overarching process layout, as the
departments are logically grouped (e.g. intensive care, nursing units, administration) whilst at a
department level there may be more of a fixed-position layout to the hospital (e.g. operation room). In
an operation room, the patient will be stationary while nurses, doctors/specialists, and equipment are
brought to the patient. Product layouts can also exist in the case of a cafeteria or lab. However, the
main concern of this study pertains to the location of the departments in relation to each other, with
the interior of the departments not being as important. Therefore, the process layout is considered
further.
66
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
When hospitals are organised by function, isolated departments exist such as laboratories, radiology,
and the intensive care unit, with each department behaving like a functional silo. Departments are
seen as singular areas whereby equipment that is similar is co-located. The entire department
specialises in similar processes and thus seeks to maximise their efficiency and utilisation (Karvonen,
Korvenranta, Paatela & Seppala, 2007). Although each patient (job or customer) may follow an
altogether different route through the system, some of the paths are more common than others, e.g.
hospitals group together certain functions, for instance emergency medicine, intensive care, radiology,
and surgery, even though they are in separate departments. This ideal arrangement would then allow
a patient entering through the emergency room to be seen in radiology, possibly surgery, and then
intensive care, and another to be admitted directly for elective surgery and then to intensive care. The
strengths and weaknesses of a process layout are shown in Table 22. The functional silos cause
complex patient flow routes and can lead to poor overall process control and thus a reduced quality of
care. This can lead to long transfer distances that involve many multi-department visits and ultimately
long patient throughput times. Scheduling is also not an easy task in such a layout.
TABLE 22: STRENGTHS AND WEAKNESSES OF A PROCESS LAYOUT ADAPTED FROM GROOVER (2007), KARVONEN ET AL. (2007), AND ALI (2015)
Strengths
Weaknesses
Ability to handle a variety of processing requirements
Not particularly at risk to equipment failures
Equipment used is less costly
Possible to use individual incentive plans
Requires lower initial capital investment
High degree of machine utilisation
Overhead costs are relatively low
Breakdown of one machine does not disturb the
production process
Supervision can be more effective and specialised
Greater flexibility of resources
In-process inventory costs can be high
Challenging routing and scheduling
Equipment utilisation rates are low
Material handling slow and inefficient
Complexities often reduce span of supervision
Special attention for each product or customer
Accounting, inventory control and purchasing are
more involved
Material handling costs are high due to backtracking
More skilled labour is required resulting in higher cost
Work in progress inventory is high needing greater
storage space
More frequent inspection is needed which results in
costly supervision
In the next section layout procedures are discussed for designing layouts, including Muther’s
Systematic Layout Procedure which is a useful process layout tool. This method is applied in
Section 3.4.
3.3 Layout Procedures
There exist three main traditional approaches to solving the FLP, namely: Apple’s Plant Layout
Procedure, Reed’s Plant Layout Procedure, and Muther’s Systematic Layout Planning Procedure
(Mehrotra, Syal & Hastak, 2005). All three layout procedures provide an overall process to construct
or improve a layout. However, these methods are very dependent on the opinion and spatial skills of
the human designer (Tompkins et al., 2010). There are also two older layout procedures in literature
67
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
called Immer’s Basic Steps developed in 1950 and Naddler’s Ideal System Approach which was
developed in 1961 (Bhatwadekar, Kulkarni & Thakur, 2015).
Some of the steps used in layout procedures are similar to variables used in the layout models
discussed in Chapter 2. For example the variable 𝑓𝑖𝑘 (the flow of material between facility 𝑖 at location
𝑘 used in the QAP resembles the flow between activities used in Apple’s Plant Layout Procedure as
well as Muther’s Systematic Layout Planning Procedure. Another example is the variable 𝑟𝑖𝑗 (the
closeness rating indicating the desirability of locating facility 𝑖 adjacent to facility 𝑗) used in the GTP
which correlates to the activities relationship chart used in Apple’s Plant Layout Procedure and
Muther’s Systematic Layout Planning Procedure. However, many of the steps of all three layout
procedures are merely guidelines for the designer and it is unlikely that two designers will produce
the same layout for a given application, even if they were using the same procedure. In contrast to this,
the algorithmic approaches (layout models) will produce the same result if they are solved optimally.
3.3.1 APPLE’S PLANT LAYOUT PROCEDURE
Apple’s Plant Layout Procedure consists of a system of 20 ordered steps as shown in Figure 13. This
method suggests that no two layout procedures are the same and therefore neither are (nor should be)
the procedures for designing them (Tompkins et al., 2010). Furthermore, an understanding of the
impact each of the steps has on each other is necessary before a holistic solution can be determined,
requiring some jumping back and forth. It is an iterative process of continuous improvement.
1. Procure the basic data 2. Analyse the basic data
3. Design the productive 4. Plan the material flow
process
pattern
5. Consider the general
material handling plan
6. Calculate equipment
requirements
7. Plan individual
workstations
8. Select specific
material handling
equipment
9. Coordinate groups of
related operations
10. Design activity
interrelationships
11. Determine storage
requirements
12. Plan service auxiliary
activities
13. Determine space
requirements
14. Allocate activities to
total space
15. Consider building
types
16. Construct master
layout
17. Evaluate, adjust, and
check the layout with the
appropriate persons
18. Obtain approvals
19. Install the layout
20. Follow up on
implementation of the
layout
FIGURE 13: APPLE'S PLANT LAYOUT PROCEDURE ADAPTED FROM TOMPKINS ET AL. (2010)
Take note of Steps 4, 10 and 13 shown in Figure 13. Step 4 refers to the flow of material. The analysis
of activity interrelationships (Step 10) is a popular approach that is also used in Muther’s Systematic
Layout Planning Procedure. Step 13 is addressed in Section 4.2.2.
3.3.2 REED’S LAYOUT PROCEDURE
Reed’s Plant Layout Procedure recommends a 10-step ‘systematic plan of attack’ for both the planning
and preparing of the layout (Tompkins et al., 2010). This is shown in Figure 14. The procedure uses a
layout planning chart which Reed considers one of the most crucial phases in the layout process. This
68
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
chart incorporates the addressing of standard times for each operation, machine selection and balance,
manpower selection and balance, as well as material handling requirements.
2. Determine the process
1. Analyse the product(s)
required to manufacture
to be produced
the product
6. Establish minimum
aisle widths
7. Establish office
requirements
3. Prepare layout
planning charts
4. Determine
workstations
5. Analyse storage area
requirements
8. Consider personnel
facilities and services
9. Survey plant services
10. Provide for future
expansions
FIGURE 14: REED'S LAYOUT PROCEDURE ADAPTED FROM TOMPKINS ET AL. (2010)
The flow of material to create a product is once again analysed in Step 2. However, no other activity
relationships are considered. Space requirements are investigated in Steps 5, 6, and 7.
3.3.3 MUTHER’S SYSTEMATIC LAYOUT PROCEDURE
Muther’s Systematic Layout Planning (SLP) Procedure is a popular process layout tool (Section 3.1)
that provides the foundation on which numerous layout design techniques and also software is
developed (Welgama & Gibson, 1995). It is based primarily on an activity relationship chart which
provides the input data, roles and relationships between activities, and material flow analysis (i.e. a
from-to chart).
Flow of
materials
Activity
relationships
Relationship
diagram
Space
requirements
Space available
Space relationship
diagram
Modifying
considerations
Practical
limitations
Develop layout
alternatives
Evaluation
FIGURE 15: MUTHER'S SYSTEMATIC LAYOUT PROCEDURE ADAPTED FROM TOMPKINS ET AL. (2010)
The steps of this method are shown in Figure 15. The concepts of Muther’s SLP Procedure are still very
popular and prominent in textbooks. It is a very simple method and generating a layout solution is
relatively easy (Tompkins et al., 2010).
69
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
3.3.4 CONCLUSION: LAYOUT PROCEDURES
In conclusion, the layout procedures do not provide a formal process or algorithm for many of the
crucial steps associated with developing a facility layout, unlike the layout models discussed in
Chapter 2. These layout models are able to help the layout analyst develop or improve a layout, while
simultaneously providing him or her with objective criteria to facilitate the evaluation of various
layout alternatives that emerge in the process. Another disadvantage of layout procedures is that they
will rarely give an optimal or near optimal solution since they are very much dependant on the
designer’s opinion and experience.
On the other hand, layout procedures are able to produce a layout with little effort within a short time.
The comparison between layout procedures and layout models suggests that it may be possible to
combine some of the steps of Muther’s SLP Procedure with a layout model (for instance both methods
analyse the flow between departments). It seems that the main advantage of using Muther’s SLP
Procedure is that it addresses the importance of placing some departments close to each other and
separating certain ones (Steps 1, 2 and 3). This approach differs from the layout models discussed in
Chapter 2 (with the exception of the GTP) which are based on the flow of material between
departments. In the next section the first few steps of Muther’s SLP Procedure are applied to the
context of a hospital.
3.4 Department level analysis
A hospital consists of a wide range of services and functional units (i.e. departments). Functional units
within a hospital oftentimes exhibit competing priorities and needs. Individual preferences and
idealised scenarios must realistically be balanced against actual functional needs (flow and
relationships to other departments), compulsory requirements, and the financial capabilities of the
hospital (Carr, 2011). A good hospital design effectively integrates functional requirements with the
human needs of its varied users (Philippi, 2012). In the context of a hospital the users include patients,
visitors, staff, suppliers, and owners.
This section applies the first few steps of Muther’s SLP Procedure, i.e. relationships between
departments. In doing this, one would need to address which hospital departments should be placed in
close proximity to each other and which ones should be placed apart.
First, existing relationship diagrams for hospitals were researched. The following key references were
found:


Pierdait (2006) developed a reference guide for hospitals (unpublished). Pierdair considered nine
hospital departments and used a scale with three ratings, namely short and required connection, short
and useful connection, and advised connection;
Kobus (2008) developed a relationship diagram with 25 departments. The diagram includes less
common units such as safety respiratory care, central processing, environmental services, maintenance,
70
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS


public facilities, and oncology. In this diagram only a few relationships were indicated (less than one
third). A scale of three ratings was used, namely essential, important, and desirable;
Motaghi, Hamzenejad, Riahi and Soheili Kashani (2011) used heurists to design a block layout for the
Shafa Hospital. They constructed a relationship chart for this specific hospital which has five floors. They
used a total of 22 units in their relationship diagram. However, there are five separate units for wards.
Installations, management, conference hall, and pavilion are also included. Furthermore, the operation
rooms are divided into two separate units. A scale with seven ratings was used, namely absolutely
necessary, special importance, important, normal, unimportant, undesirable, and quite undesirable; and
Neufert and Neufert (2012) propose a relationship chart for 11 departments. They used a scale of three
ratings, namely, very good connection required, good connection sensible, and connection is desirable.
Thus no generic relationship diagram for a district hospital was found and for this reason a new one is
developed in this section. First, the function of each department of a district hospital was researched
and its relationships with other departments were determined. Three of the references found used a
scale with three ratings for relationships between departments. Motaghi et al. (2011) used the most
comprehensive scale for relationships since their scale had seven ratings and included negative
relationships. It was therefore decided to use this scale for rating the relationships between
departments of a district hospital. In order to show these ratings more clearly, each rating is assigned
to a numerical value between -10 and 10 as shown in Table 23.
TABLE 23: CLOSENESS RATING SCALE FOR DEPARTMENTAL RELATIONSHIPS
Departmental relationship
Closeness rating
Absolutely necessary
10
Especially important
7
Important
5
Desirable
3
Unimportant
0
Undesirable
-3
Important separation
-5
Absolutely necessary separation
-10
As mentioned in Section 1.7 different types of hospitals exist, namely district hospitals, regional
hospitals, provincial tertiary hospitals, central hospitals, and specialised hospitals. According to
Conradie and Steyn (2014) the service package for South African hospitals can be divided into
outpatient services, inpatient services, clinical support services, administrative support services, and
facilities management services. The departments in each field are shown in Figure 16.
71
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS



Kitchen
Laundry
Central stores


Administration
Patient
admissions unit
General ward
Medical ward
Surgical ward
 Paediatrics
 Obstetrics
 Intensive care
unit
 High care unit
 Psychiatric
ward



Clinical support services
Emergency unit
Outpatient unit
Day surgery
Inpatient services


Administrative support services

Facilities management services
Outpatient services
HOSPITAL SERVICE PACKAGE
Radiology
Operating
theatres
 Rehabilitation
 Sterile supply
unit
 Pharmacy unit
 Mortuary


FIGURE 16: SOUTH AFRICAN HOSPITAL SERVICE PACKAGE
According to the Regulations Governing Private Hospitals and Unattached Operating Theatre Units
(further discussed in Section 4.2.2) the following departments are relevant to district hospitals in
South Africa, namely Administration (Admin), Obstetrics (Obst), Operating Suite (OS), Paediatric Unit
(Paed), Laundry, Kitchen, Sterilisation and Disinfection Unit (S&DU), Pharmacy, Emergency and
Casualty Unit (Emer), Acute Psychiatric Facility (Psych), Chronic Care Unit (CCU), Rehabilitation Unit
(Rehab), Mortuary, Laboratory, Radiology, Neonatal Intensive Care Unit (NICU), Intensive Care Unit
(ICU), and High Care Unit (HCU). All of these departments are now analysed in order to develop a
generic relationship diagram for district hospitals.
3.4.1 ADMINISTRATION
The administration department is defined by Broekmann and Steyn (2014) as “offices for
administrative staff, clinicians, matrons, and other medical staff; board rooms, auditoria, conference
rooms, chapel, library, staff rest rooms, ablutions and sanitary facilities; facilities for safekeeping of
goods, stationery, stores, and cleaners’ rooms.” The administration department manages daily
operations of all departments in the hospital through planning, directing, co-ordinating, and
supervising the delivery of healthcare. Functions of this unit include planning, implementing, and
evaluating the following: departmental activities; facility utilisation studies; daily operation of patient
admissions and clinical facilities; health information and medical records, and; maintenance and
security of all patient records. This unit also implements health policies set by government as well as
recruit, hire, and train medical, administrative, and other technical staff (Gupta, Gupta, Kant,
Chandrashekhar & Satpathy, 2007).
The administration department should be orientated in a way that is easily accessible to the public, but
that is at the same time private and secured with access control (Broekmann & Steyn, 2014). Offices
for
hospital
management
may
be
located
in
more
private
areas
away
from
the
entrance (Kunders, 2004). Financial areas that have a functional relationship with the public must be
72
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
close to the entrance of the hospital, e.g. all outpatients come to the administration department to
register and pay their account. It is therefore necessary to place the outpatient department in close
proximity to the administration department. Outpatients refer to patients who seek diagnosis or
treatment at a hospital, but are not admitted for an overnight stay while inpatients refer to patients
who are admitted (Medical Dictionary for the Health Professions and Nursing, 2012).
The main entrance of the hospital (part of the administration department) typically leads into the
central spine of a hospital. From here visitors are directed to the inpatient services, e.g. obstetrics,
paediatrics, and patients are directed to clinical support services, e.g. radiology.
The department relationships with the administration department are therefore suggested to be as
follows: Obst(3), OS(3), Paed(3), Out(10), Laundry(0), Kitchen(0), S&DU(3), Pharmacy(3), Emer(0),
Psych(3), CCU(3), Rehab(3), Mortuary(0), Laboratory(0), Radiology(4), NICU(3), ICU(3), and HCU(3).
3.4.2 OBSTETRICS
Obstetrics deals with childbirth, but it is practically impossible to separate from pregnancy
(maternity) related problems. General hospitals customarily combine obstetrical and gynaecological
services. Gynaecology on the other hand, treats disturbances and diseases peculiar to women. The
obstetric unit is responsible for providing safe and efficient care that ensures utmost safety and
comfort for women and new-borns (Kunders, 2004). The nursing unit of this department provides
prenatal care, observation, and comforting of the patients in labour, providing assistance in the
delivery room, care of the mother after delivery and also care of the new-born (Department of
Health, 2002).
The maternity care of a hospital typically spreads over two departments (Van der Schyf &
Flemming, 2015). The antenatal services and postnatal services are usually provided in the outpatient
department of a hospital while the antenatal ward, high-dependency unit, delivery unit, postnatal
ward, well-baby nursery, kangaroo mother care, and obstetric theatres are located in the obstetrics
department. For this reason, a close relationship with the outpatient department is required. The
obstetrics department must be adjacent to the operating suite (unless it has a dedicated operating
theatre) and the neonates’ (new-borns) ward (Van der Schyf & Flemming, 2015). Therefore, high
relationships ratings are assigned to the operating suite and the paediatrics department (since the
neonates’ ward is within this unit). Some hospitals design the delivery suite adjacent to the surgical
suite. An advantage of such design is that the two departments require the same isolation, type of
nursing service, cleaning, air-conditioning, and sterile supplies. However, a disadvantage of this design
is inter-traffic and possible cross-contamination. Furthermore, close proximities to the HCUs and ICUs
are preferable. Easy access to both the pharmacy and radiology is required.
73
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
The clinical section of this department should be located in a secluded area. Within the clinical
facilities, the labour and delivery suites should be as remote as possible as well as easily accessible
from the entrance to the department. Furthermore, the nursing unit has a close relationship with the
pharmacy, laboratory, and the kitchen. This department is usually noisy and it is ideal to separate it
from the rest of the hospital (Kunders, 2004).
The suggested department relationships with the obstetrics department are therefore as follows:
Admin(0), OS(10), Paed(10), Laundry(3), Kitchen(3), S&DU(3), Pharmacy(7), Emer(3), Psych(-3),
CCU(-3), Rehab(-3), Mortuary(1), Laboratory(3), Radiology(7), NICU(10), ICU(9), and HCU(9).
3.4.3 OPERATING SUITE
The operating suite department is defined by Van Reenen (2014) as “a highly controlled environment
for the operative and perioperative care of patients who are undergoing diagnostic and surgical
procedures under anaesthesia.” The surgical facilities include the operating rooms, duty station,
recover area, change rooms, storage, and setting up area. In a typical general hospital surgical patients
represent 50 to 60 percent of the admissions, and a high percentage of them undergo surgery
(Kunders, 2004). There are three types of surgical patients, namely inpatients, outpatients, and
ambulatory surgery patients. Inpatients are those who are hospitalised for surgery. Outpatient surgery
relates to usually minor surgical procedures that require local anaesthesia and the patients are
admitted and discharged the same day (Kunders, 2004). About 55% of all surgical procedures are
done on an outpatient basis (Wier, Steiner & Owens, 2015). The terms ambulatory surgery (also
known as same-day surgery) and outpatient surgery are often incorrectly used interchangeably.
Ambulatory surgery usually requires more extensive procedures, e.g. getting your wisdom teeth
removed (Whitlock, 2015).
The best location for this department is one which allows efficient flow of patients, staff, and clean
supplies (Kunders, 2004). Ideal adjacencies include emergency, radiology, laboratory, intensive care
unit, sterilisation and disinfection department, and delivery suite (obstetrics). The sterilisation and
disinfection department are responsible for preparing and autoclaving all surgical instrument, gloves,
linen, syringes, and needles. This department may be adjacent, but it should not be placed adjacent to a
theatre room in order to prevent cross-contamination (Van Reenen, 2014). A relatively lower rating is
thus assigned to this relationship. Easy access from the emergency department and delivery suite are
crucial. The ICU should be adjoined. It is also important to separate the operating suite from the rest of
the hospital. Entry and exit into this department should be controlled. There should be no traffic
through it and no interference from other departments. It is ideal to place this department away from
the main traffic and other possible sources of noise, e.g. administration and outpatient departments.
Lastly, it must also be accessible from wards.
74
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
The suggested department relationships with the operating suite are therefore as follows: Admin(-2),
Obst(10), Paed(0), Out(-2), Laundry(0), Kitchen(0), S&DU(5), Pharmacy(0), Emer(10), Psych(3),
CCU(3), Rehab(0), Mortuary(3), Laboratory(8), Radiology(8), NICU(7), ICU(10), and HCU(7).
3.4.4 PAEDIATRIC UNIT
The paediatric unit caters for children up to 12 years of age (neonates, infants, and
children) (Flemming, 2014). The care includes preventative, promotive, curative (assessing, classifying
and treating), and rehabilitative care (Department of Health, 2002). As much healthcare as possible is
provided in the home, and children are admitted and kept in hospital only when this is essential. Up to
15 percent of beds in a general hospital may be required for paediatric patients (Kunders, 2004). An
important design consideration is to acknowledge the psychological role that parents and friends play
in the recovery and rehabilitation of the sick child. The Paediatric Unit is generally noisy and should
therefore be located away from the main stream of hospital traffic. It is ideal to place this department
close to the main entrance, but separate from the inpatient facilities (wards) in a way that children
from the paediatric unit can go home without having contact with the other wards (Flemming, 2014).
The neonatal nursery of the paediatric unit should be located close to obstetrics since that is where the
maternity delivery unit is located and the postnatal wards. Similar to other nursing units, a close
relationship with the operating suite, pharmacy, central stores, laboratory, and the dietary (kitchen)
are required (Kunders, 2004).
The department relationships with the paediatric unit are therefore as follows: Admin(3), Obst(8),
OS(5), Laundry(5), Kitchen(5), S&DU(3), Pharmacy(5), Emer(3), Psych(-5), CCU(-3), Rehab(-3),
Mortuary(3), Laboratory(5), Radiology(3), NICU(10), ICU(-2), and HCU(-3).
3.4.5 OUTPATIENT UNIT
The outpatient department provides medical care that does not require an overnight stay in a hospital
(known as ambulatory care) (Flemming, 2014). Ambulatory care involves preventative, promotive,
curative, and rehabilitative services. Activities in the outpatient department include patient triage,
consulting, counselling, examination, observation, diagnosis, treatment, and therapy. As mentioned
earlier, the outpatient department provides antenatal and postnatal care for mothers and
unborn/born child. Close proximity with obstetrics (and paediatrics) are therefore ideal.
It is important to place the outpatient department close to the main entrance (administration) since
there are constant movement between these two units (Broekmann & Steyn, 2014). Another reason is
that patients’ medical records need to be readily accessible from the outpatient department.
Furthermore, easy access to the rehabilitation unit and other wards are required.
It is essential that the pharmacy is close to the outpatient unit because many patients attend the
hospital pharmacy to collect dispensed medicines after consultations or treatment (Flemming, 2014).
75
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
Good access to the radiology department is necessary since patients often attend the unit during the
course of an outpatient session. This department is usually noisy since patients, staff, service staff, and
escorts are constantly moving throughout the consulting rooms, to treatment rooms, service points,
and other departments. It is useful to place this department away from relatively quiet units, e.g. the
operating suite.
The department relationships with the outpatient unit are therefore as follows: Admin(10), Obst(10),
OS(-2), Paed(8), Laun(3), Kitchen(0), S&DU(3), Pharmacy(10), Emer(3), Psych(3), CCU(5), Rehab(5),
Mortuary(3), Laboratory(3), Radiology(5), NICU(3), ICU(3), and HCU(3).
3.4.6 LAUNDRY
The function of a hospital’s laundry is to receive contaminated items for cleaning and providing an
adequate, economic, continuous supply of clean, disinfected linen to all patient care units in a
hospital (Fourie, Sheared & Steyn, 2014). The laundry services are one of the most important
supportive services in a hospital, i.e. washing/cleaning of linen (Gupta et al., 2007). Depending on the
hospital policy, it may be undertaken in-house or it may be outsourced. The main functions of a
hospital laundry are to supply a clean and adequate amount of linen on a regular basis, and procedures
must be taken to prevent cross infection (Natarajan, 2010). Positive relationship ratings are thus
assigned to the wards, emergency department, operating suite and outpatient department. As a
planning guide, the quantity of linen to clean per day may be approximated as 4kg per bed, where up
to 20% is infected linen. In addition, 8kg of soiled linen for each surgical operation and delivery should
be considered. The laundry typically has a separate entry and exit which helps to separate clean and
soiled linen areas (Gupta et al., 2007). The hospital laundry should be located away from the main
service and traffic of the hospital (Fourie, Sheared & Steyn, 2014).
The department relationships with the laundry are therefore as follows: Admin(0), Obst(4), OS(4),
Paed(4), Out(3), Kitchen(0), S&DU(0), Pharmacy(0), Emer(4), Psych(4), CCU(4), Rehab(3),
Mortuary(0), Laboratory(0), Radiology(0), NICU(4), ICU(4), and HCU(4).
3.4.7 KITCHEN
The food service department of a hospital is an important supportive service which provides
inpatients their meals as per their dietary requirements, taking into consideration the nature, type of
disease, nutritional requirements, and food habits of the patient. The food requirements of staff as well
as visitors are catered for too. The department may be outsourced, depending on the hospital’s
policy (Gupta et al., 2007). It should be located centrally to supply patients’ needs within the shortest
time (Steyn & Boltman, 2014). Ideally, it should be placed close to wards, but it must be located so that
heat and odours are not directed towards traffic areas. As a fire safety precaution they should also not
be located under wards, especially those for ambulant patients (Kunders, 2004). It should be located
76
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
away from normal hospital activities, but trucks should be able to easily deliver supplies such as meat,
vegetables, and dairy products (Natarajan, 2010).
The department relationships with the kitchen are therefore as follows: Admin(-3), Obst(3), OS(-3),
Paed(3), Out(0), Laundry(0), S&DU(-3), Pharmacy(-3), Emer(-3), Psych(3), CCU(3), Rehab(3),
Mortuary(0), Laboratory(0), Radiology(0), NICU(3), ICU(3), and HCU(3).
3.4.8 STERILISATION AND DISINFECTION UNIT
The primary function of the sterilisation and disinfection unit is providing efficient, economic, quality
supply of sterilised items to all patient-care points, and to receive contaminated items for
cleaning (Steyn & Sheard, 2004). In some hospitals, this department also purchases, stocks, and
distributes supplies. It is essential that sterilisation methods, procedures and process are carried out
at a central department in order to maintain quality, standardisation, cost-effectiveness, control, and
optimal utilisation (Gupta et al., 2007). The sterilisation and disinfection unit should be centrally
located with access to all service areas such as the ICU, wards, and emergency department (Manav,
2011). It should have direct access to the operating suite (Steyn & Sheard, 2004).
The department relationships with sterilisation and disinfection unit the are therefore as follows:
Admin(0), Obst(7), OS(10), Paed(7), Out(3), Laundry(0), Kitchen(0), Pharmacy(0), Emer(7), Psych(3),
CCU(7), Rehab(5), Mortuary(0), Laboratory(0), Radiology(0), NICU(7), ICU(7), and HCU(7).
3.4.9 PHARMACY
The hospital pharmacy supplies and dispenses necessary drugs and medical supplies. It purchases
drugs from an identified supplier to maintain adequate quantities, as well as receives, records, and
stores them (Department of Health, 2002). Traffic within this department should be flexible and
economical. Provision must be made for securing dangerous drugs and bulk quantities should be kept
elsewhere (Kunders, 2004).
It is important that the pharmacy is located in a way that it is accessible to the outpatient department,
and convenient for dispensing whilst maintaining patients’ privacy and dignity (Kunders, 2004). It is
often the patient’s last stop within any hospital establishment visit (thus a relationship with the
administration department). Pharmaceuticals are also distributed to the inpatient areas under the
supervision of the pharmacist. Delivery to the pharmacy should also be considered. The pharmacy
should ideally be placed central to the core services near outpatients, inpatients, and the emergency
department (Nice, 2014).
The department relationships with the pharmacy are therefore as follows: Admin(5), Obst(3), OS(3),
Paed(3), Out(10), Laundry(0), Kitchen(0), S&DU(0), Emer(10), Psych(3), CCU(8), Rehab(3),
Mortuary(0), Laboratory(0), Radiology(0), NICU(3), ICU(3), and HCU(3).
77
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
3.4.10
EMERGENCY AND CASUALTY UNIT
The emergency and casualty unit is defined by Fleming (2014) as “the dedicated area in a health facility
that is organised and administered to provide a high standard of emergency care to those in the
community who are in need of acute or urgent care.” The emergency department is an essential unit of a
hospital, and must be available to receive patients 24 hours a day. Its purpose is to manage patients
with a wide variety of critical, urgent and semi-urgent condition. These patients are received, triaged,
and stabilised (Haryana State Health Resource Centre, 2013).
When planning the emergency department, one should understand that the word ‘emergency’ is often
misunderstood, and that patients often seek emergency care for a situations other than critical medical
conditions since patient’s perception of an emergency can vary (e.g. a person who is unable to contact
a doctor at night to treat his constipation that has lasted a few days may feel like an emergency). Thus,
the emergency department provides for a comprehensive range of services, from first aid and general
outpatient services to sophisticated surgical and medical emergencies and full-scale trauma care,
e.g. treatment and reporting of physical and psychological abuse (Department of Health, 2002).
Therefore, it is important to incorporate flexibility in the design to handle a range of cases
economically and efficiently (Kunders, 2004). In the case of a disaster in a region, the emergency unit
should be able to manage patients who are victims (Haryana State Health Resource Centre, 2013;
Department of Health, 2002).
It is vital that the emergency department has rapid access to the operating suite, and radiology
department (over 40% of emergencies require X-rays and portable X-rays are usually not
satisfactory) (Kunders, 2004). A separate entrance, away from the main hospital entrance, is required
and should be easily accessible to ambulances and patients. Traffic control is critical since injured
patients, victims, and their families can cause inefficient operation and consideration should be given
to protection of privacy (Haryana State Health Resource Centre, 2013). The laboratory should be
accessible from the emergency department since a sizable number of patients require laboratory
tests (Kunders, 2004). Additionally, smooth access and proximity to the ICU and HCU in order to
minimise the transfer times of critically ill patients (Fleming (2014). The emergency department
should have 24-hour access to the hospital pharmacy. The mortuary should also be accessible. The
emphasis of these relationships should be on rapid access.
The department relationships with the emergency and casualty unit are therefore as follows: Admin(3), Obst(5), OS(10), Paed(3), Laundry(0), Kitchen(0), S&DU(3), Pharmacy(7), Psych(0), CCU(3),
Rehab(-3), Mortuary(4), Laboratory(9), Radiology(10), NICU(5), ICU(5), and HCU(5).
3.4.11
ACUTE PSYCHIATRIC FACILITY
The Acute Psychiatric Facility provides medical and nursing support for patients in periods of acute
psychiatric illness (National Health Service Confederation, 2012). The care provided involves a full
78
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
range of disciplines, including psychologists, pharmacists, and occupational therapists. While some
mental illnesses are treated at specialised mental hospitals, a large number of people receive
psychiatric diagnosis and treatment of varying degrees at general hospitals. The care of these patients
requires
knowledge
of
their
various
behavioural
patterns
and how
to
best
manage
them (Kunders, 2004).
In order to provide care for acute psychiatric patients, laboratory tests (e.g. basic blood tests and HIV
testing) or services for additional investigations are often required. Radiographic services are also
often utilised, e.g. to exclude trauma (Department of Health, 2002). It is important that the psychiatric
unit is protected and in a secure seclusion area which is under close permanent supervision of
staff (Kunders, 2004). It should ideally be located away from the other hospital departments and have
a separate entrance. In order to prevent self-harm, single-storey buildings are preferred. Another
consideration is to create a therapeutic environment and give attention to noise levels, colour, natural
light, and spaces that would promote a healing environment (Railoun & Van der Schyf, 2014). It is thus
suggested to locate this department away from the main traffic areas (e.g. Admin), noisy departments
(e.g. Paed, Emer) and critical units (e.g. ICU).
The department relationships with the acute psychiatric facility are therefore as follows: Admin(-3),
Obst(-3), OS(-3), Paed(-3), Out (-3), Laundry(3), Kitchen(3), S&DU(0), Pharmacy(3), Emer(-3), CCU(0),
Rehab(0), Mortuary(-4), Laboratory(6), Radiology(4), NICU(-3), ICU(-3), and HCU(-3).
3.4.12
CHRONIC CARE UNIT
The Chronic Care Unit (CCU) provides medical care for adults in all life stages with long term illnesses,
as opposed to acute care which is concerned with emergency treatment and critical care (e.g. ICU or
HCU) (Department of Health, 2002). The CCU combines patient bed space and clinical treatment space
which allows the unit to provide an effective, economical, and therapeutic unit for patients (Van der
Schyf & Fleming, 2014). This unit is complex in the sense that it may extend over a prolonged period of
time and requires input from different health professionals, various medications, and possibly
monitoring equipment. The support of family and friends usually plays a big role and should be
considered in planning the department. It is important that noise is deduced, patient safety maximised,
and the work environment aesthetically pleasing.
The CCU and other inpatient units (e.g. Paed, Obst) form the core of a hospital and have relationships
with most departments, especially operating suite, pharmacy, central stores, and laboratory. The
administration department should have a primary circulation route to the CCU. There should also be a
reasonable relationship with the mortuary, laundry, and kitchen (Van der Schyf & Fleming, 2014).
79
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
The department relationships with the chronic care unit are therefore as follows: Admin(5), Obst(0),
OS(5), Paed(0), Out(5), Laundry(5), Kitchen(5), S&DU(3), Pharmacy(5), Emer(5), Psych(0), Rehab(0),
Mortuary(5), Laboratory(5), Radiology(3), NICU(0), ICU(5), and HCU(5).
3.4.13
REHABILITATION UNIT
The purpose of a rehabilitation unit is to manage acute and chronic disabling conditions, prepare the
disabled person for reintegration into the community as well as to provide training, assisting and
counselling activities on a daily basis. This includes tasks such as bathing, dressing, eating, and
communicating (Amrpa, 2016). This unit also manage those problems related to disability that: cannot
be managed in the community, support rehabilitation services in the community, and supply
appropriate assistive devices to patients. Patients are also taught to service and maintain the assistive
devices (Department of Health, 2002). Consideration should be given to the fact that many patients
treated in these units use wheelchairs or walking aids. Types of patients typically include post-surgery
recovery, multiple trauma, amputees, and neurological injuries and disease (Buhrs, 2013).
For the rehabilitation unit it is important that patients get enough sleep and rest. High noise levels or a
sudden increase of noise can prevent patients getting restful sleep which may negatively impact their
rehabilitation process due to the side effects of sleep deprivation, e.g. impaired memory, learning and
well-being (Yelden, Duport, Kempny & Playford, 2015). For this reason, noisy departments such as
obstetrics, the paediatric unit, and the emergency department should be located away from the
rehabilitation unit. Another consideration is that family members form an integral part of the
rehabilitation team and are encouraged to be present during treatments (Buhrs, 2013). Thus,
promoting and accommodating visitors in this unit is of importance (relationship with Admin). Similar
to other inpatient units, a reasonable relationship with the kitchen, pharmacy, and laundry is
suggested.
The department relationships with the rehabilitation unit are therefore as follows: Admin(5), Obst(-5),
OS(5), Paed(-5), Out(5), Laundry(5), Kitchen(5), S&DU(3), Pharmacy(5), Emer(-5), Psych(3), CCU(3),
Mortuary(3), Laboratory(3), Radiology(3), NICU(3), ICU(3), and HCU(3).
3.4.14
MORTUARY
A hospital’s mortuary holds dead bodies until burial can be arranged. It also provides a place where
relatives and other people can view and identify bodies. The needs of visitors and patient dignity in
handling bodies should thus be considered in the design (Reenen, 2014). Visitors should be provided
with easy access to the mortuary upon arrival, without having to travel unnecessarily through other
hospital departments. A separate entrance could thus be ideal. The mortuary should be located that it
is easily accessible to mortuary staff and related service providers without presenting emotional or
ethical problems for unrelated staff, visitors or patients (Kunders, 2004). When bodies are moved,
they should not be moved through the main traffic areas. It is thus suggested to locate the mortuary
80
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
away from Admin and Psych. Pathologists are sometimes required to investigate causes of death and
make scientific investigations. Ideally, the mortuary should be located near the pathology department
or laboratory. In order to move bodies to the mortuary it should be accessible from the wards (Paed,
Obst,
CCU,
NICU,
ICU,
HCU),
emergency
department
(and
outpatients)
and
operating
suite (Natarajan, 2010).
The department relationships with the mortuary are therefore as follows: Admin(-3), Obst(4), OS(4),
Paed(4), Out(4), Laundry(0), Kitchen(0), S&DU(0), Pharmacy(0), Emer(4), Psych(-3), CCU(4),
Rehab(2), Laboratory(5), Radiology(0), NICU(4), ICU(4), and HCU(4).
3.4.15
LABORATORY
A hospital laboratory performs tests in order to enable staff to make or confirm diagnoses as well as
the treatment and prevention of diseases (Kunders, 2004). Laboratory tests are either specialised or
routine, e.g. urine analysis and blood cell counts. The primary design objective of a laboratory should
be to create environments in which laboratory activities can be conducted without compromising
patient dignity, laboratory process or health and safety risks (Van Reenen, 2014). It should be located
in a way that they cannot be used as thoroughfares for staff, patients or visitors. However, patients
should
have
direct,
suitable
access
and
chemicals
should
easily
be
delivered
and
collected (Kunders, 2004). The laboratory serves the emergency, outpatient, and admitting
departments. It should also be close to radiology and obstetrics. Since test specimens are transported
from inpatient wards and clinical support services, reasonable relationships with these departments
are expected.
The department relationships with the laboratory are therefore as follows: Admin(3), Obst(5), OS(5),
Paed(5), Out(3), Laundry(0), Kitchen(0), S&DU(3), Pharmacy(0), Emer(9), Psych(6), CCU(5), Rehab(5),
Mortuary(0), Radiology(4), NICU(5), ICU(5), and HCU(5).
3.4.16
RADIOLOGY
The main role of the radiology department is to assist in the diagnosis and treatment of
diseases (Department of Health, 2002). In large hospitals, this service may be arranged into three
departments, namely: diagnostic radiology, therapeutic radiology, and nuclear medicine. Usually
medium and small-sized hospitals only have diagnostic radiology available. Its responsibilities include
taking, developing, and interpreting X-rays. It also conducts research and participates in educational
programs for nurses, technicians, and the community (Kunders, 2004).
It is crucial that the radiology department is easily accessible to the emergency department, outpatient
department, and inpatient wards. The department should be placed central to these areas and take
into consideration convenience, privacy, and traffic flow (Kunders, 2004). Other important
relationships include the ICU and operating suite. It is possible to use potable radiographic equipment
81
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
in selected instances for imaging of patients, e.g. ICU. It should also be noted that the devices in the
radiology department are large and bulky and pose challenges to install. Access is not only important
during
construction,
but
also
during
maintenance,
repair
and
replacement
of
these
devices (Coetzer, 2013). A closer relationship to the entrance is thus suggested.
The department relationships with the radiology are therefore as follows: Admin(5), Obst(3), OS(5),
Paed(3), Out(7), Laundry(0), Kitchen(0), S&DU(0), Pharmacy(0), Emer(10), Psych(4), CCU(3),
Rehab(3), Mortuary(0), Laboratory(0), NICU(3), ICU(3), and HCU(3).
3.4.17
NEONATAL INTENSIVE CARE UNIT
The Neonatal Intensive Care Unit (NICU) specialises on the health of neonates who require constant
nursing,
difficult
surgical
procedures,
continual
respiratory
support,
or
other
interventions (Kunders, 2004). A neonate is defined as a child from birth to one month of age (or a two
month old child weighing less than 2kg). Neonates are very vulnerable patients and consideration to
the design of the wards is therefore essential. Quiet times in NICUs are very important since loud noise
levels have an impact on neonates’ ability to absorb oxygen. It is recommended that NICUs have a
separate area in the hospital in order to minimise the transmission of diseases and maintaining
privacy (De Jager, 2014). Yet, the unit should be well integrated with the other hospital departments
for easy access (Bird, Bostic, Taylor & Zhou, 2011). It should also have ready access to the maternity
unit (Obst), emergency unit, operating suite, and laboratory. Since the paediatric department
accommodates more stable neonates, a close relationship with this department is suggested.
The department relationships with the NICU are therefore as follows: Admin(3), Obst(10), OS(8),
Paed(10), Out(3) Laundry(3), Kitchen(0), S&DU(5), Pharmacy(5), Emer(5), Psych(0), CCU(0),
Rehab(0), Mortuary(3), Laboratory(5), Radiology(3), ICU(0), and HCU(0).
3.4.18
INTENSIVE CARE UNIT
The Intensive Care Unit (ICU) can be defined as a “highly specialised and technically sophisticated
dedicated unit for the management and care of critically ill patients who are dependent on invasive life
support, and intensive levels of medical and nursing care that requires complex treatment” (Coetzer &
Fleming, 2014). This care differs from other hospital units since ICU patients require a higher level of
observation and monitoring and may have special equipment in their room, e.g. ventilators, heart
monitors, respiratory monitors (Sutter Health, 2014).
The ICU should be located in a distinct area within the hospital preferably with controlled access. It
should be centrally located with no thoroughfare through the unit. Supply and staff traffic should be
separated from visitor traffic and no through traffic to other departments should occur. It is important
that the unit is close to the emergency department, obstetrics (maternity unit), operating room (OS),
radiology department, intermediate care units (e.g. HCU), recovery rooms, and respiratory
82
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
therapy (Society of Critical Care Medicine, 1995). It also requires access to the hospital pharmacy and
laboratory. The specialised cardiac team should be able to respond promptly to an ICU emergency call.
Admissions to ICUs are through the emergency department or from the operating rooms after major
surgery. It should also not be too far from the inpatient units since this will reduce the traveling
distance and time to move patients between the two departments, especially in the case of an
emergency to move a patient from a ward to the ICU (Kunders, 2004). Patients are typically moved to
the HCU or inpatient wards, if the patient’s condition improves.
The department relationships with the ICU are therefore as follows: Admin(3), Obst(9), OS(10),
Paed(4), Out(3), Laundry(3), Kitchen(3), S&DU(5), Pharmacy(5), Emer(10), Psych(0), CCU(5),
Rehab(3), Mortuary(5), Laboratory(5), Radiology(7), NICU(0), and HCU(9).
3.4.19
HIGH CARE UNIT
A High Care Units (HCU) is where patients are cared for more extensively than in a normal ward, but
not to the point of ICUs (Coetzer & Fleming, 2014). The relationships with the HCU are thus rated
between the ICU and CCU relationships. Patients in these units are not on life support systems. This
unit is appropriate for patients after major surgery, for those with single-organ failure, patients who
are at risk of requiring intensive care admission, or as a step-down between ICUs and ward-based
care. It is usually located adjacent to the ICU (Intensive Care Society, 1997).
The department relationships with the HCU are therefore as follows: Admin(3), Obst(3), OS(7),
Paed(5), Laundry(3), Kitchen(3), S&DU(5), Pharmacy(5), Emer(5), Psych(0), CCU(7), Rehab(3),
Mortuary(4), Laboratory(5), Radiology(5), NICU(0), and ICU(10).
3.4.20
RESULTING DEPARTMENTAL RELATIONSHIPS AND FLOWS
The department relationships often have different ratings, e.g. on the one hand you want the
paediatric unit (Paed) to be close to the ICU (4 assigned) in case a paediatric patient goes into a critical
condition. But on the other hand, the paediatric department is usually noisy and may disturb critically
ill patients from getting rest (-2 assigned). For cases like this, a score between the two values are used
(2 assigned). The resulting relationship diagram is shown in Figure 17.
83
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
Administration
Obstetrics
Operating Suite
Paediatric Unit
Outpatient Unit
Laundry
Kitchen
Sterilisation and DisinfectionUnit
Pharmacy
Emergency and Casualty Unit
Acute Psychiatric Facility
Chronic Care Unit
Rehabilitation Unit
Mortuary
Laboratory
Radiology
Neonatal Intensive Care Unit
Intensive Care Unit
High Care Unit
3
3
10 3
10 10
4 10 0
-2 5 -2
8 5 3 0
5 0 5 5
3 3 8 5 -3
0 5 2 5 -3
0 3 4 10 -3 5
0 10 3 -2 -2 4
0 0 3 -4 4 -4 -3
0 0 0 -2 3 3 4
0 -3 3 5 -4 4 5 5
7 3 5 5 3 7 5 3
3 0 4 4 3 5 7 10 3
3 5 4 0 3 3 8 9 3
-3 6 3 0 0 5 10 10 8
4 4 0 0 0 3 2 7
0 -3 0 1 0 4 3 2
3 4 0 0 2 4 3
1 -3 9 0 6 3 4
4 6 10 4 6 3
3 5 4 4 4 6
3 4 -3 4 4
3 3 0 -2 4
0 1 5 -2
2 3 3 5
5 4 3
3 5 4
5 5
0 5
0
10
FIGURE 17: RELATIONSHIP DIAGRAM OF HOSPITAL DEPARTMENTS
The flow diagram shown in Figure 18 was generated based on the research of each department as well
as consulting Carr (2011), Tarawneh (2014), and the World Health Organization (2000).
Facilities management
services
Inpatient services
Paed
Obst
Kitch
CCU
Rehab
Laun
Psych
HCU
ICU
NICU
Clinical support services
Mort
Lab
Administrative support services
Admin
Main
Entrance
Outpatient services
OS
Radio
Emer
Emergency
Entrance
Out
Outpatient
Entrance
Phar
FIGURE 18: FLOW DIAGRAM OF HOSPITAL DEPARTMENTS
84
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
3.5 Hospital design considerations
Hospital planning has not historically incorporated designs that are aimed at improving quality of
service and safety of patients, yet significant funds are invested in healthcare facilities
annually (Hughes, Reiling & Murphy, 2008). Studies have mainly focused on the effects of light, colour,
views, and noise, yet there are many more considerations that affect the quality of care in
hospitals (Barach & Dickerman, 2007). Therefore, an investigation was conducted into the most
important design considerations applicable to hospital layouts found in literature. The results involve
considerations revolving around: patient-centeredness, efficiency, flexibility and expandability,
sustainability, and therapeutic environments, as shown in Table 24.
TABLE 24: QUALITATIVE HOSPITAL DESIGN CONSIDERATIONS
Design objectives
Description
Patient-centeredness
Patients’ need and wants are understood and
addressed, including patient safety and
infection prevention
Efficiency
Concerns the rate at which medical care is
provided to patients, including:



Lean management
Occupant flows
Ergonomic factors
Flexibility and
expandability
The ability to change infrastructure,
technology, and management structure
Sustainability
Endurance of the hospital over a lengthy
period of time
Physical setting and organizational culture
supporting patients and families, including:
Therapeutic
environment





Noise
Windows vs no windows
Sunny rooms
Multiple occupancy vs single patient rooms
Supportive design
References
Shrewsbury and Telford Hospital
NHS Trust (2014)
Life Healthcare (2014)
Hughes et al. (2008)
University Hospitals Institute (2013)
Carr (2011)
Hughes et al. (2008)
University Hospitals Institute (2013)
Carr (2011)
Shrewsbury and Telford Hospital
NHS Trust (2014)
Life Healthcare (2014)
Hughes et al. (2008)
Carr (2011)
Ferenc (2015)
Barach and Dickerman (2007)
Carr (2011)
Each of the design objectives is now further discussed.
3.5.1 PATIENT-CENTEREDNESS
Hospitals are built for the primary reason of providing medical care for patients. Patient-centeredness
involves not only medical care but also non-medical needs and wants, such as patient safety (e.g. the
protection against hospital acquired infections). The following design elements were found to promote
patient-centeredness (Hughes et al., 2008):

Using variable-acuity rooms and single-bed rooms;
85
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS

Ensuring sufficient space to accommodate family members;

Enabling access to healthcare information; and

Having clearly marked signs to navigate the hospital.
Hospitals exhibit several particular safety concerns in addition to the general concerns of other
buildings. These concerns involve the protection of hospital property and assets (including drugs),
protection of patients (including incapacitated patients) and staff, prevention against the spread of
infections, and safe control of violent or unstable patients (Carr, 2011). Designing for patient safety
includes the following: choosing the correct materials for walls, ceilings, glass, and doors; using the
correct hardware such as hinges, closers, and locksets; installing fire extinguishing systems; choosing
the correct furniture (refer to Hunt and Sine (2013) for a lengthy discussion of each). These
considerations are not included in the layout framework of this study since they are not within the
study’s scope.
There exist different ways in which to prevent the transmission of infections: using ventilation and
filtration systems, choosing surfaces that can easily be decontaminated, and facilitating hand washing
with the availability of sinks and alcohol hand rubs (Institute of Medicine, 2013). According to Zimring,
Joseph and Choudhary (2004) hospital-acquired infections (caused by pathogens) are among the
leading causes of death in the United States, resulting in the deaths of more people than AIDS,
automobile accidents, or breast cancer. A research team identified more than 120 studies linking
infection to the built environment of the hospital. Infections are transmitted through contact or are
even airborne. The literature suggests that there is a clear relationship showing that infection rates are
lower when there is higher air quality and patients are kept in single-bed as opposed to multi-bed
rooms. Furthermore, several studies reveal that single-bed rooms also lessen the risk of infection due
to contact, again promoting the advantage single-bed rooms (Barach & Dickerman, 2007). This is due
to the fact that environmental surfaces and features are easily contaminated around infected patients
via contact with both patients and/or staff. In comparison to single-bed rooms, multi-bed rooms are
much more difficult to decontaminate properly once a patient is discharged, increasing the difficulties
posed by multiple surfaces acting as infection reservoirs. Thus, it can confidently be said that singlebed rooms are superior to multi-bed rooms with regards to reducing transmission of infections
through the air or via contact (Institute of Medicine, 2013).
3.5.2 EFFICIENCY
Efficiency in the context of a hospital relates to providing quality medical care with the least waste of
time and effort. Eliminating waste through continuous improvement is termed lean management. This
involves flowing the product at the pull of the customer (Bhuiyan & Baghel, 2005). The product in the
context of a hospital is the cured, treated, and/or diagnosed patients. This flow through a hospital has
a significant impact on the overall efficiency. Furthermore, a hospital’s productivity can be
86
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
significantly improved by taking ergonomic factors into account. Each of these areas will now be
addressed.
Lean management
Lean management aims to increase the value of the customer with fewer resources and in order to
achieve this, the focus shifts from optimising separate departments to optimising the flow of products
and services that add value to these areas. The five main principles of lean are (Leaning
Forward, 2005):

Determine what creates value from the customers’ perspective;

Identify all the steps of the process chain;

Ensure the flow of processes;

Produce only according to the needs of customers; and

Aim for perfection by continuously eliminating waste.
In the context of a hospital the patient is the main focus and all the processes that improve (add value
to) the patient’s health should be optimised. All processes that do not add any value to the patient’s
health is waste. Waste in the context of a hospital is defined as follows (Centre for Special Studies and
Programs, 2010):

Overproduction: when medication is given out early, treatments done in order to balance the
utilisation of staff, and the duplication of tests;

Transportation and motion: when the same patient, specimens, supplies, or workers are moved;

Waiting times: include patients waiting for bed assignments, discharge, or testing results;

Processing: includes the duplication of procedures, retesting, and unnecessary paperwork;

Inventory: consists of specimens waiting for analysis, pharmacy stock and supplies; and

Defects: includes medication errors, missing information, wrong procedure taken, or wrong patient
treated.
All of these wastes increase costs and take unnecessary time from staff that could have been used to
treat patients. Since healthcare needs to be affordable to all of society, it is important to minimise
waste as far as possible. However, the quality of services should be maintained. Thus cost should be
minimised while maintaining quality service, minimising errors, and reducing waiting times. The
layout of the hospital determines how far the patients need to travel to get treated at the various
hospital departments. Since transportation of patients does not add to the health of patients, it can be
seen as a non-value adding activity. In fact, Huang and Irani (2013) found that transportation causes
anxiety for patients, is a risk of injury to patients, and can lead to spreading of infection to or from the
patient. When patients travel between hospital departments there is a delay in their treatment
process. This delay could be life-threatening for patients in critical conditions. Queues can form in
corridors, elevators, and transport between buildings, causing further delays. The total time taken to
complete one patient’s treatment thus increases. There can also be a cost involved regarding
87
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
personnel hired to transport the patients. Karvonen, Korvenranta, Paatela and Seppala (2007) found
that the transportation distance from a clinical unit to its internal provider has a negative relationship
with the quality of care provided by the staff as well as a positive relationship with the duration the
patient spends in the healthcare system (Soriano-Meier, Forrester, Markose & Arturo
Garza-Reyes, 2011).
Occupant flow
There exist different types of occupant flows in a hospital such as patient flows, nurse flows, doctor
flows, visitor flows, and supporting personnel flows. The effective flows of each can mean the
difference between life and death for patients. The waiting times of patients are one of the six priority
areas of the National Core Standards (discussed in Section 4.2.3 and Appendix B). The transportation
of patients throughout the hospital can be seen as a non-value adding activity since it does not
contribute to the end product, i.e. cured, treated, or diagnosed patient. Transportation involves travel
delays between locations, increases total time taken to complete patient treatment, and the cost of
personnel hired to transport. In addition to this, it causes anxiety for patients, increases the risk of
injury to patients, and adds to the transmission of infections.
A general rule for designing a hospital is to separate patient, visitor, and staff flows. The use of
dedicated resource corridors is particularly important in preventing congestion and delays (Rechel,
Wright, Barlow & McKee, 2010). Flows of goods, patients, and work also need to be separated, so as to
enable each to move according to the logic and pace most suitable for each. Focus should therefore be
placed on similar processes rather than on similar clinical conditions. A good example is that of an
emergency department in an Australian training hospital whereby patients were divided into two
streams according to complexity rather than urgency, thus creating a fast-track patient stream for
patients who can be treated and discharged very quickly which maximised throughput (Saghafian,
Hopp, Van Oyen, Desmond & Kronick, 2012). This resulted in significant improvements in several
other key performance indicators as well, including mean waiting time, and treatment time.
It is important to differentiate between continuous and batch processes since failing to distinguish can
lead to seeing queuing as a lack of capacity (in terms of beds, diagnostics, doctors, facilities, or nurses).
Investments aimed at increasing the capacity of a hospital often fail since they are not systematically
directed at the most pertinent areas, i.e. the bottlenecks. There are often feedback loops, hidden
bottlenecks, and lines moving at varying speeds, all of which negatively impact performance. Thus
insufficient supply may well also be a problem, but it can only be understood as a function of how the
service is (or rather should be) configured (Rechel et al., 2010). The emergency department, intensive
care unit, and operating rooms as well as their related pre- and post-care areas are more often than
not the most important bottlenecks to deal with, due to the fact that they are non-interchangeable
resources.
88
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
Ergonomic factors
Ergonomic designs are aimed to promote efficiency and ease of use. Ergonomic factors include
displays, layout of panels and machines, seating, thermal comfort, noise, and lighting (Konz &
Johnson, 2004). Many of these topics are not within the scope of this study, yet the layout of a hospital
can be designed to promote ergonomic efficiency by designing aisles with adequate widths for its
purpose (e.g. an aisle may adhere to the minimum requirements for aisle widths but it might not be
practical for patients and staff).
TABLE 25: NON-CRITICAL AND CRITICAL MINIMUM PASSING SPACES ADAPTED FROM THE ACCIDENT COMPENSATION CORPORATION (2011)
Minimum passing spaces
Non-critical passing (m)
Critical passing (m)
Patient beds
1,8
2,2
Large patient handling equipment
1,5
1,8
Wheelchair passing space
1,5
1,8
Patients assisted by carers
1,5
1,8
The main considerations for an aisle include where it is located, how frequently it is used by staff and
patients, the equipment that is used, and whether the passing of people and equipment is critical or
non-critical (Hall-Andersen & Broberg, 2014). Table 25 shows various considerations for corridors.
Critical passing is where the unrestricted movement is important, and usually applies to emergency
evacuation routes and corridors that have a high frequency of usage. On the other hand, non-critical
passing is where immediate clear passage for patients are not critical, and usually applies to corridors
with low-frequency use (Accident Compensation Corporation, 2011).
3.5.3 FLEXIBILITY AND EXPANDABILITY
Since medical needs and modes of treatment change continuously, it is important for hospitals layouts
to design for flexibility and expandability (Life Healthcare, 2014). The layout should thus be treated as
dynamic. Just like businesses are wise to have business strategies that are long-term, hospitals should
ideally have a similar long-term master plan for how the layout may need to change to remain
operating at an optimal level. This master plan must ideally be consistent with the hospital’s business
plan, and go as far as possible to anticipate future requirements in advance, making provisions for
when it must adapt to changing hospital requirements (Shrewsbury and Telford Hospital NHS
Trust, 2014). For these reasons, the following design considerations are proposed (Tompkins
et al., 2010):

Follow modular concepts of space planning;

Use generic room sizes as much as possible instead of highly specific ones;

Be open-ended, with thoroughly planned directions for future expansion;

Minimise layout size to avoid wasted time and motion of workers;

Eliminate centralized storage and move storage to various departments; and
89
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS

Minimise the amount of reorganisation that will be made necessary by future growth and change.
In addition to the design considerations mentioned, technology and management structure should also
be able to adapt to future requirements.
3.5.4 SUSTAINABILITY
Sustainability relates to the endurance of the hospital over a lengthy period of time (Hughes
et al., 2008). This involves using financial resources as efficiently and effectively as possible. Hospitals
have a significant impact on both the economy and environment of the community surrounding them,
being heavy users of energy and water, and generating relatively large quantities of waste. Because of
these demands on community resources, sustainable design is a critical consideration. The benefits of
sustainable
design
include
reduced
operational
costs,
and
a
better
therapeutic
environment (Carr, 2011). Examples of such designs would for instance maximise utilisation of natural
daylight, optimise acoustic performance, and incorporate appropriate ventilation and moisture
control (Carr, 2011).
3.5.5 THERAPEUTIC ENVIRONMENT
The importance of the environment for a patient’s health and well-being dates back as far as 400BC
with Hippocrates, and the 19th century with Florence Nightingale (Huisman, Morales, Van Hoof &
Kort, 2012). Ulrich (2001) found scientific evidence that certain environments in a hospital can
promote improved patient medical outcomes while others can worsen patient health. 80% of the most
rigorous studies have positive links between environmental characteristics and patient health
outcomes (Ulrich, 2001). Huisman et al. (2012) conducted a similar study on 798 papers and found
that 50% of the papers provide strong evidence for this link, with 86% of them finding a link between
the interactions of patients and their families with health outcomes. Furthermore, they found that the
well-being of healthcare staff are also impacted by the nature of the environment, but evidence of staff
outcomes is scarce and insufficiently substantiated. Environmental factors that have an effect on
health outcomes of patients include noise levels, presence of windows, sunlit spaces, occupancy rate,
and supportive design.
Noise levels
Noise levels in hospital are typically more than 15 to 20 dB higher than those recommended by WHO
(Basner, Babisch, Davis, Brink, Clark, Janssen & Stansfeld, 2014). Thus, hospital noise may be a threat
to patient rehabilitation and staff performance. Noises in hospitals, especially in intensive care units,
are characterised by irregularly occurring noises from sources such as medical devices, telephones or
pagers, conversations, door sounds, and nursing activities. High noise levels were found to negatively
affect patients in the following ways (Basner et al., 2014): decreased rate of wound healing; higher
levels of disturbance and annoyance; decreased oxygen saturation, elevated blood pressure, increased
heart and respiration rate; sleep disruption and awakening; higher incidence of rehospitalisation, and;
90
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
cognitive impairment (mainly in children). Similarly, noise levels have the following effects on staff
(Joseph & Ulrich, 2007): increases their perceived work pressure, stress, and annoyance; increases
fatigue; emotional exhaustion and burnout, and; difficulty in communication which potentially leads to
making more errors.
The layout of a hospital can play a role in reducing noise levels by implementing the following
designs (Herman Miller Inc., 2006; Joseph & Ulrich, 2007):

Providing single-patient bedrooms: single rooms minimise noise caused by visitors, traffic, and that of
care givers attending to other patients;

Removing or reducing loud noise sources: such as washing and drying units, and other equipment and
machinery that is used in caring for patients;

Providing private rooms enclosed with walls that go up to the ceiling: the use of curtains alone is less
than optimal for reducing noise transfer which is better effected through utilising solid walls;

Enclosing examination and treatment areas with walls: similarly, noise cannot travel as well through
solid walls;

Providing private discussion areas in admitting areas as well as on the unit for private conferences with
families and staff: this allows families to console one another and grieve without interrupting treatment
of other patients or their recovery process;

Increasing the distance between sound sources and people (sound intensity decreases by 6 dB every
time this distance is doubled): location is key to effectively managing noise levels and must be
considered prior to building a hospital; and

Decentralising nurse stations: this disperses people which reduces the concentration of sound
emanating from their activities.
Presence of windows
Studies on critical-care patients were conducted on rooms with and without windows and notable
evidence of negative effects were found for the latter (Hughes et al., 2008). Studies have linked the
absence of windows with heightened rates of depression, anxiety, and delirium relative to rates for
similar patients in rooms with windows (Joint Commission Resources, 2003). It was also found that
the employees who work in rooms with nature viewing windows have less stress, better health status
and higher job satisfaction.
Sunlit spaces
An investigation was conducted in a Canadian hospital with patients with severe depression and it was
found that patients assigned to rooms overlooking sunny spaces had, on average, a shorter hospital
stay than the patients assigned to rooms overlooking spaces in shadow/gloom (Zimring et al., 2004).
Another study found that the mortality numbers of patients with myocardial infarction decrease when
they are assigned to rooms overlooking sunny spaces (Ulrich, 2001). A study that used questionnaires
also shows that the employees prefer workplaces with window views of spaces illuminated by
91
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
sunshine rather than shadows. However, rooms exposed to direct sunlight can create bright glare
patches which affect patients and employees negatively.
Occupancy rate
The main argument for single rooms over multiple occupancy rooms is that the infection rates are
lower (Chaudhury, Mahmood & Valente, 2004). However, the initial cost per bed for single-occupancy
rooms is higher for acute care units. Some even found that multiple occupancy provide each of the
patients with helpful social support (Ulrich, 2001). On the other hand, conflict among roommates can
lead to costly room changes and patient moves may over the lifetime of the hospital exceed the initial
construction cost advantages for multiple occupancy rooms (Calkins & Cassella, 2007). Several studies
found that the presence of other patients in multiple-occupancy rooms can be a major source of stress,
mainly due to a loss of privacy (Ulrich, 2001).
Supportive design
Supportive design refers to environmental characteristics that facilitate coping and restoration for the
stress that accompanies illness and hospitalisation (Fischl, 2006). Supportive healthcare environments
can foster gains in patient health outcomes. The following guidelines are proposed for creating
supportive healthcare environments (Ulrich, 2001):

Foster control: give patients a real/perceived sense that they are in control of their circumstances and
determine what others do to them in order to help them deal better with stress and improve their health.
Patients can feel more in control if they have, for example, sufficient information, and control over eating and
sleeping times. The design of single rooms can promote this as well as improve patient safety (Hughes et
al., 2008). Other design approaches include providing privacy, gardens accessible to patients in wheelchairs,
architectural design that promotes wayfinding in large hospitals, and privacy in imaging areas. Employees
can feel a sense of control when they have comfortable break rooms and easily adjustable workstations since
it allows them to briefly escape from workplace demands and stressors;

Foster social support: encourage and support the presence of family and friends. This can be achieved by
providing comfortable waiting areas, convenient access to rest rooms and food, attractive gardens with
sitting areas that facilitate socialising with patients, and even convenient overnight accommodations; and

Provide access to nature and other positive distractions: prolonged exposure to nature views can help
calm patients as well as improve other health outcomes. One study found that a bedside window overlooking
trees had more favourable recovery courses than patients overlooking a brick building wall. Another study
found that patients have less anxiety and required fewer strong pain doses if they were exposed to a nature
picture compared to no picture or an abstract picture (Kellert, Heerwagen & Mador, 2011).
3.5.6 LINKING DESIGN CONSIDERATIONS TO QUANTITATIVE METHODS
Thus far in this chapter, the qualitative design methods have been analysed, the manner in which they
relate to the quantitative design methods analysed in Chapter 2 has been shown (Section 3.3), and a
number of hospital design considerations have been discussed. The aim of this next section is to
92
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
illustrate how the hospital design considerations relate to the quantitative methods presented in
Chapter 2.
The five hospital design considerations with their corresponding measures as well as how they relate
to the layout models, as discussed in Chapter 2, are shown in Table 26.
TABLE 26: LINKING HOSPITAL DESIGN MEASURES AND LAYOUT MODELS
Flexibility and
expandability
Sustainability
x
x
x
x
x
x
Minimise waiting times
Minimise cost
x
x
Minimise travel distances
Minimise costs
x
Linear Mixed Integer
Model
x
x
Linear Continuous
Model
Mixed Integer
Programming Problem
Optimise occupant flows
Minimise transportation
Design measures
Graph Theoretic
Problem
Linear Integer
Programming Problem
Efficiency
Quadratic Set Covering
Problem
Design
objectives
Quadratic Assignment
Problem
Layout model
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Table 27 shows, some of the design considerations that are not part of the standard model
formulations, but are included in the framework as constraints (Section 4.2.2) or can be included
through additional constraints, e.g. a therapeutic environment can be promoted in a hospital by adding
single rooms and private discussion areas in the layout design.
TABLE 27: INCORPORATION OF HOSPITAL DESIGN METHODS IN FRAMEWORK
Design
objectives
Patientcenteredness
Efficiency
Flexibility and
expandability
Sustainability
Design measures
Variable-acuity rooms (private rooms)
Space for family members
Single occupancy rooms
Separate patient, visitors and staff flows
Design critical passing spaces
Design dedicated resource corridors
Follow modular concepts
Use generic room sizes
Minimise layout size
Decentralised storage
Maximise daylighting
93
Incorporated in
framework as
constraints
x
x
x
Can be incorporated
through additional
constraints
x
x
x
x
x
x
x
x
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 3 | QUALITATIVE LAYOUT DESIGN METHODS AND HOSPITAL DESIGN CONSIDERATIONS
INCORPORATION OF HOSPITAL DESIGN METHODS IN FRAMEWORK (CONTINUED)
Therapeutic
environment
Single occupancy rooms
Enclosed examination and treatment
areas
Provide private discussion areas
Design patient rooms with windows
(adjacent to outside)
Design patient rooms to overlook sunny
spaces (correct orientation)
Multi occupancy rooms
Adjacent to gardens
Staff break rooms
Provide waiting areas for family members
with access to rest rooms and food
x
x
x
x
x
x
x
x
x
3.6 Conclusion
Qualitative design methods and hospital design considerations were presented in this chapter. Links
with the quantitative design methods of the previous chapter were also illustrated. Furthermore, the
first steps of Muther’s SLP Procedure were applied to a hospital context to generate a relationship
diagram for a set of hospital departments. The key findings are summarised in the paragraphs that
follow.
It was found that hospitals are classified as hybrid layouts, but the arrangement of its departments is
seen as a process layout. In a process layout, organisations are organised by function and its units
behave like functional silos.
Three main traditional approaches to solving the FLP were found and analysed, namely: Apple’s Plant
Layout Procedure, Reed’s Plant Layout Procedure, and Muther’s SLP Procedure. These layout
procedures were found to be vague and rarely provide an optimal / near optimal solution since they
are very much dependent on the designer’s experience and opinion. However, it was suggested that
some of the steps of Muther’s SLP Procedure (a process layout tool) can be combined with a layout
model analysed in Chapter 2.
An investigation was conducted into the most important design considerations applicable to hospital
layouts found in literature. The results involve considerations revolving around: patient-centeredness,
efficiency, flexibility and expandability, sustainability, and therapeutic environments. The way in
which these considerations relate to the layout models presented in Chapter 2 is determined.
With a foundation in place regarding quantitative and qualitative layout design methods, it is now
necessary to understand the context of a rural hospital. Thus, Chapter 4 investigates the differences
and similarities between urban and rural hospitals, including the numerous constraints applicable to
the various hospital rooms.
94
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4
RURAL VERSUS URBAN HOSPITALS
“The all-pervading disease of the modern world is the total imbalance between city and countryside, an
imbalance in terms of wealth, power, culture, attraction and hope. The former has become over-extended
and the latter has atrophied. The city has become the universal magnet, while rural life has lost its
savour.”
Schumacher (2000)
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
95
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
4.1 Introduction
In this chapter, the differences and similarities between rural and urban hospitals are investigated.
Rural and urban communities differ from one another in certain aspects, for example: resource
availability, access to medical care, and rural specific illnesses and attitudes towards health. The
question is one of whether this influences the needs of a hospital layout for a rural community. To
answer this question a comparison is made between the two types of hospital settings, and thereafter
deductions are made regarding the implications for the floor layout. The main objectives of a rural
hospital layout are identified.
In order to design a hospital that is suitable to serve a rural community, it is important to understand
the health needs and challenges specific to rural areas whilst also drawing from the regulations and
standards common to both urban and rural hospitals, as shown in the chapter structure of Figure 19.
COMMONALITIES AMONG
4.2
Building regulations and
standards
RURAL AND URBAN
HOSPITALS
Quantitative
considerations
4.3
RURAL-SPECIFIC
Healthcare personnel
constraints
Argument for
layout model
objectives
Rural community
considerations
CONSTRAINTS
Qualitative
consideration
Infection prevention
factors
Single rooms
FIGURE 19 RURAL VERSUS URBAN COMPARISON STRUCTURE AND OBJECTIVES
Before investigating rural hospitals, it is necessary to discern what a rural hospital is. Alternative
terms found in literature for a rural hospital are ‘district hospital’, ‘community hospital,’ and ‘general
hospital’ (Jamison, Breman, Measham, Alleyne, Claeson, Evans, Mills & Musgrove, 2006). Hospitals can
be divided into three levels, and rural hospitals fall under the first level, i.e. primary-level hospital care
(discussed in Section 1.7). This means there are oftentimes only limited laboratory services available
for pathological analysis, and few specialists, with the most typical ones providing internal medicine,
obstetrics, paediatrics, and general surgery (Jamison et al., 2006).
4.2 Commonalities among rural and urban hospitals
Rural hospitals and urban hospitals have many similarities regarding general building constraints,
such as the National Building Standards Act of 1977, the South African National Standard 10400,
Regulation 158 of the South African law, and hospital adherence rules which are governed by the
National Core Standards. Each of these will be unpacked in the following sections, so as to better
understand the basic premises on which buildings layouts (both rural and urban) are structured.
96
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
4.2.1 GENERAL BUILDING CONSTRAINTS
All buildings in South Africa are required to adhere to the National Building Standards Act 1977.
According to Section s. AZ4, 1(b) of this act, all buildings also need to comply with the requirements of
the relevant part of the South African National Standard 10400.
The National Buildings Standards Act 1977 s. C1, 1 requires that each room or space of any building
should have dimensions that will ensure that said room or space is fit for the purpose for which it is
intended. The minimum dimensions for each department and facility of a hospital are laid out in
Regulation 158 and will be taken into account in the next section.
The National Buildings Act 1977 Section s. S1, 1 requires that facilities must accommodate persons
with disabilities. This has an effect on the hospital layout in terms of necessary aisle widths and the
design of bathroom facilities.
4.2.2 HOSPITAL LAYOUT CONSTRAINTS
Hospitals have specific constraints that they need to adhere to, e.g. certain departments and rooms are
compulsory. Each department and room also needs to be bigger than a certain size according to a set
criterion. Therefore, this section examines the minimum area for each hospital room according to the
size of the hospital (i.e. number of hospital beds) as well as determine which rooms and departments
are optional. According to the Department of Health (2002), district hospitals usually have between 30
and 200 beds. The Department of Health is the custodian of healthcare in South Africa regarding care
delivered at either public or private healthcare facilities. However, the responsibility is entrusted to
the provincial departments.
According to De Jager (2011), the Regulations Governing Private Hospitals and Unattached Operating
Theatre Units (otherwise known as Regulation 158, or more succinctly R158) are still in use for
governing the design and operations of hospitals in South Africa, having been developed in 1980 and
last updated in 1993. The Western Cape developed and gazetted their provincial regulations, R187 in
2001 and modified it in 2003. The Eastern Cape and Gauteng both prepared draft provincial
regulations, but neither have been finalised or formally approved. According to Baloyi (2011) those
developed by Gauteng appear to be the most comprehensive and it is proposed that these be accepted
as a discussion document and put out for comment with a view to being developed for national use.
Since the provincial standards are not accepted as the national standards yet, this study will focus on
the requirements set out in R158.
Therefore, all the constraints of the R158 were analysed first. However, this document does not specify
or recommend all the sizes of each room, for which reason additional regulations were incorporated,
including constraints provided by the KwaZulu-Natal Department of Health (2010), as well as
constraints that are clearly set out in other countries (e.g. Scotland and the United Arab Emirates).
97
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
Additionally, articles, books, and industrial norms were researched. Appendix A provides a detailed
outline of this analysis. The minimum dimensions for each room are shown, as well as the relevant
references and information regarding which rooms are optional. The following hospital departments
were analysed: Administration, Obstetrics, Operating Suite, Paediatric Unit, Laundry, Kitchen,
Sterilisation and Disinfection Unit, Pharmacy, Emergency and Casualty Unit, Acute Psychiatric Facility,
Chronic Care Unit, Rehabilitation Unit, Mortuary, Laboratory, Radiology, Neonatal Intensive Care Unit,
Intensive Care Unit, and High Care Unit.
4.2.3 ADHERENCE TO STANDARDS
National Department of Health developed the National Health Insurance (NHI) policy in order to
improve health. The NHI aims to ensure that everyone has access to quality healthcare regardless of
their socio-economic status. This will lead to changes in management systems, service delivery and
administrative systems. In preparation for implementation of the NHI, service delivery in South
African healthcare facilities needs improvement. Thus a national quality assurance program called the
National Core Standards (NCS) was launched to drive facility improvements and serve as a benchmark
for quality. All South African facilities, both in the private and public sectors, shall be required to
comply with these standards in 2025 (National Department of Health, 2011:3b).
The NCS is divided into seven structured domains, namely: patient rights, patient safety and clinical
governance, clinical support services, public health, leadership and corporate governance, operational
management, and facilities and infrastructure. The seventh domain (facilities and infrastructure) of
the NCS is relevant to this study and is included in Appendix B as well as its sub-domains and
corresponding standards and criteria. This domain covers the requirements for clean, safe and secure
physical infrastructure, hotel services and effective waste disposal. The remainder of the standards do
not pertain to the construction of the hospital nor its layout, but rather the functioning of it.
Standard 7.1.3 requires that waiting areas are convenient and provide adequate shelter and seating for
patients. The corresponding criterion 7.1.3.1 states that the waiting areas should be appropriately
located and adequate for the number of patients using them. Therefore, for areas of the layout where
queues will most likely form, waiting areas should be provided.
Standard 7.5.1 requires that waste management in the health establishment complies with legal
requirements, national standards, and good practice. With regards to the hospital layout, waste
management is taken into account in Section 4.1.2 by including waste room(s) in the layout. Other
necessary rooms included are linen and laundry services of Standard 6.6.1 and food services of
Standard 7.7.1.
98
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
4.3 Rural-specific constraints
Rural and urban lifestyles, health, and illnesses differ in many ways. According to
Wainer and Chesters (2000), the negative factors associated with rural lifestyles include: social
stresses, exacting behavioural norms expected within the community, limited modes of transportation
and its low availability resulting in high travel costs, and limited work opportunities leading to
unemployment and poverty. Furthermore, the rural environment is characterised by distance, specific
occupational hazards, uncontrolled environments, sparse infrastructure, and risk taking attitudes
towards health, illness and behaviour (Humphreys, 1998). Rural inhabitants are not protected against
adverse conditions and circumstances (e.g. droughts, fires, floods, and recessions) as are urbanites
(Wainer & Chesters, 2000). The aim of this section is thus to identify hospital layout design objectives
that are specific to rural communities.
4.3.1 HEALTHCARE PERSONNEL CONSTRAINTS
One vital resource which affects the manner in which a hospital’s departments must be arranged
pertains to the number of healthcare personnel assigned to the hospital. Healthcare personnel scarcity
in rural regions impacts the flow requirements of the establishment. Furthermore, due to rural
communities typically being deep in poverty, their ability to pay for basic procedures, never mind
advanced ones, determines to a large extent the types of facilities necessary, as some will quite simply
never be used, thus having an impact on which rooms are necessary.
According to the National Department of Health (2011), 43.6% of South Africa’s population live in
rural areas. The latest provincial rural population percentage breakdown is given by Kok and
Collison (2006) as shown in Table 28. The percentage medical scheme coverage of each province as
estimated in 2011 by the South African Health Review (2013) is also shown in Table 28. It is clear that
the majority of South Africans do not have a medical aid plan, regardless of whether they live in rural
or urban areas. However, it is also evident that the more rural provinces (e.g. Limpopo) exhibit even
lower coverage than more urban ones. The reason for these low statistics may be due to the average
South African’s lack of financial security.
TABLE 28: RURAL POPULATION SIZE AND THE CORRESPONDING MEDICAL SCHEME COVERAGE OF PROVINCES IN SOUTH AFRICA ADAPTED FROM NATIONAL
KwaZulu-Natal
Limpopo
Mpumalanga
Northern Cape
North West
Western Cape
South Africa
Medical scheme coverage (%)
Gauteng
Rural population (%)
Free state
Measurement
Eastern Cape
DEPARTMENT OF HEALTH (2011)
62
25
4
55
90
61
20
59
10
42
10.5
14.4
27.3
12.1
7.9
15.9
16.6
15.4
24.7
16.9
99
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
According to Wits Centre for Rural Health Strategy (2008), the provinces with the highest percentage
of rural inhabitants have the lowest number of medical practitioners per capita. This finding is
supported by Table 29 which shows the public as well as private medical practitioner count per
100,000 people in 2011. Table 29 further indicates that the total number of healthcare personnel
(i.e. human resources for health, or HRH) per 100,000 people in 2010 (with the exception of the North
West) was substantially lower in the more rural provinces. The 2008 South African Democratic and
Health Survey showed that this scarcity of nurses is more prominent in rural areas (Eygelaar &
Stellenberg, 2012). This scarcity of healthcare personnel changes the dynamics of how best these
personnel might be aided by the correct hospital layout, i.e. to maximize flow. According to the
National Department of Health (2011), the rural population in South Africa (43.6% of the national
population) is served by only 12% of the doctors and 19% of the nurses.
TABLE 29: HUMAN RESOURCES AND MEDICAL PRACTITIONERS PER PROVINCE IN SOUTH AFRICA ADAPTED FROM NATIONAL DEPARTMENT OF HEALTH
Eastern Cape
Free state
Gauteng
KwaZuluNatal
Limpopo
Mpumalanga
Northern Cape
North West
Western Cape
South Africa
(2011)
Total human resources for health per
100,000 population per province in 2010
448
520
692
588
488
452
555
331
741
557
Public medical practitioners per 100,000
population in 2011
29.
7
30.
0
40.
2
46.
9
26.
8
31.
6
53.
9
23.
0
45.
2
36.
6
Private medical practitioners per 100,000
population in 2011
17.
7
31.
0
73.
2
29.
3
9.7
21.
1
19.
7
23.
7
76.
4
37.
6
Measurement
On average, about 1,200 medical students graduate each year in South Africa and less than 3% of these
continue to work in rural areas in the long run. According to the National Department of
Health (2011), the factors which affect the lack of healthcare professionals working in rural areas
include the following: decreased funding, poor social infrastructure, fear of safety, lack of work
opportunities for spouses of health workers, deficiencies in infrastructure, lack of opportunities for
schooling children, and lack of commensurate compensation for dealing with these negative factors.
Eygelaar and Stellenberg (2012) argue that the main barriers to receiving adequate medical care in
rural areas are: the remoteness of such communities (distance to them), the high costs to access them,
and problems with accessing, supporting, and relieving qualified hospital staff.
An additional problem is that some healthcare professionals do not report for their Community
Service (CS). ‘Community service’ in this context refers to a requirement of health professionals to
complete one year of practice in the public sector following their graduation. A survey by the National
Department of Health (2009) reported that 17% of health professionals did not report for CS and a
further 6.1% reported that they would emigrate after completing their CS. This amounts to roughly
100
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
23% of health professionals planning to leave South Africa. According to the NHI Policy Proposal
(2009) the number of medical graduates entering into CS has decreased by 73.5% between 1999 and
2008 (from 1,112 to 295 per annum). This shortage of staff results in those who remain having even
more arduous working conditions to deal with, as well as the accompanying stress of it all (Wits
Centre for Rural Health Strategy, 2008). Several initiatives have been introduced to alleviate HRH
shortages, such as the rural and scarce skills allowances, CS, and the Occupational Specific
Dispensation. Therefore, it is of great importance that the design of a hospital layout is optimal, such
that it enables efficient running and utilisation of its HRH resources.
This dwindling number of healthcare staff in rural hospitals results in a high workload and an
extremely stressful work environment which subsequently leads to long patient queues and
overworked staff. Therefore, optimal usage of the medical practitioners is critical and efficiency is thus
an important design consideration. In other words, it is important to optimise objectives such as
hospital flow and travel time within the hospital.
4.3.2 RURAL COMMUNITY CONSIDERATIONS
Rural-specific challenges include gaining access to medical care, dealing with rural-specific illnesses,
and lax attitudes towards health. These are discussed in this section and are shown as a cause and
effect diagram in Figure 20.
High indirect costs of patients
Accommodation expenses
Long travel distances
Lost income due to waiting times
Rural-specific challenges
Shortage of medical supplies/equipment
Long patient queues
Low patient health outcomes
Shortage of ambulances
Overworked staff
Transportation problem
Long patient queues
Low funds to build a new facility
Shortages of staff
Low funds to renovate or expand hospital
Unpaid/inadequately paid staff
Lack of funds
FIGURE 20: CAUSE AND EFFECT DIAGRAM OF RURAL-SPECIFIC CHALLENGES FACED BY RURAL COMMUNITIES
Access to medical care
According to the South African Constitution (1996), Section 27, “every person has the right to access
health services.” More than 16.6% (5.2 million) people in South Africa experience difficulties in
accessing healthcare and those most deprived are the rural communities (Shisana, 2007). Rural
communities often do not make use of the nearest hospital and this could be due to the fact that access
101
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
to healthcare is generally costly and inconvenient (National Department of Health, 2011). For example,
the cost of traveling, paying for accommodation, time away from work and family, and operating hours
of medical facilities play a role in this. This is exacerbated by the fact that due to limited healthcare
personnel in rural areas the doctor is not always present when and if a patient makes their way to the
hospital. The biggest of these challenges is that rural residents must travel greater distances in order
to receive the necessary medical care which usually includes high indirect costs (e.g. bus tickets,
accommodation). According to Goudge (2009), poor rural households in Limpopo spend up to 80% of
their monthly income on health related costs. Although outside the scope of this study, giving
consideration to choosing a hospital location which minimises the travelling distance for rural
communities could make a significant impact on the health status of rural communities.
These challenges to accessing a healthcare facility often lead to healthcare needs being met by private
alternatives. There is some evidence that suggests illegal health clinics often operate in townships and
that people are prepared to pay for such services despite the obvious risks (Coetzer & Pascarel, 2012).
Hand in hand with higher levels of poverty come lower levels of health in rural communities when
compared to urban communities (National Department of Health, 2011). According to the
SAHRC (2007), poverty has long been recognized as a major cause of ill-health and acts as a barrier to
accessing healthcare services. From Table 30 it can be seen that the three most rural provinces have
the highest labour force unemployment rates (for those aged 15-64 years old). Jobs in rural settings
are often physically strenuous in nature, e.g. farming, fishing, forestry, mining, and manufacturing
(Strong & Elliot-Schmidt, 1997). Such professions are naturally linked to more severe health and safety
hazards as a result of longer working hours, lack of enforcement of safety measures, and usage of
makeshift tools. This results in a heightened frequency of work related injuries.
39%
2.7
1.1
3.3
4.2
2.2
South Africa
33%
Western Cape
26%
North West
Limpopo
33%
Northern
Cape
KwaZuluNatal
37%
Mpumalanga
Gauteng
Unemployment rates for labour force
aged 15-64 from Census 2011 (Statistics
SA, 2013)
Number of children in millions under 18
(General Household Survey, 2011)
Free state
Measurement
Eastern Cape
TABLE 30: UNEMPLOYMENT RATES AND CHILDREN LIVING IN POVERTY
32%
27%
32%
22%
30%
0.4
1.3
1.8
18.5
64%
31%
58%
37%
14%
35%
1.5
Percentage of children living in income
poverty* (General Household Survey,
74% 59% 34% 67% 76% 57% 63%
2011)
Percentage of children living in
households without an employed adult
51% 35% 16% 43% 50% 29% 42%
(General Household Survey 2011)
*Income poverty: Households with a monthly income per capita less than R604 in 2011
102
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
Poverty also affects the ability of a healthcare facility to operate and deliver quality healthcare
services. This lack of funds on the part of patients may in turn cause a shortage in medical supplies and
equipment, a shortage of ambulances, low funds to maintain the infrastructure, and unpaid or
inadequately paid healthcare staff.
According to Children’s Gauge (2013), the percentage of young children (those under 16 years of age)
living in rural areas is as high as 45% and as Table 30 shows, most of the children live in poverty
stricken households, i.e. households that exhibit income poverty meaning they have an equivalent
monthly income per capita of less than what R604 could buy in 2011. This high incidence of children
influences the need for crèches and facilities to look after not only unhealthy children but also those
that are too young to be left at home while a parent or guardian is confined to the hospital.
Another issue is that residents lose their income as a result of waiting in queues for many hours
(sometimes even days) at healthcare facilities (Coetzer & Pascarel, 2012). In general, it appears that
access to financial capital in general is a problem faced by the rural communities especially in the
provinces with larger rural populations. In addition, the three most rural provinces have the highest
unemployment rates as well as the highest percentages of young children to care for. For these
reasons, lack of access to financial capital and high indirect costs are major concerns for rural
communities.
Rural-specific illnesses
In the South African context, the burden of some chronic conditions such as diabetes is evenly
distributed across socio-economic groups, whilst others such as HIV/AIDS, Tuberculosis, depression,
STD, and diarrhoea have a substantially greater negative impact on lower income groups (Statistics
SA, 2013).
Personal health choices such as diet and cleanliness impact the life expectancy of residents. Rural
residents are less likely to have health problems related to poor air quality or crime. The Australian
Institute of Health and Welfare (2005) found a significant difference in that residents outside major
cities are more likely to have the following health problems: coronary heart disease and circulatory
disease, chronic obstructive pulmonary disease, type 2 diabetes, asthma, infectious diseases, poor
dental and oral health, disabilities, and prostate and colorectal cancer. Zithulele Hospital, a rural
hospital in the Eastern Cape of South Africa, identified four priority areas that cause the majority of
disease, suffering and even disability in the rural community (Jabulani Rural Health Foundation,
2014). As would be expected, the main areas are HIV/AIDS, Tuberculosis (TB), maternity related
infections, and child sicknesses.
HIV/AIDS is a prominent health concern in South Africa which has the highest prevalence of HIV
compared to any other country in the world with 5,6 million people living with HIV and 270,000 HIV
103
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
related deaths recorded in 2011 alone (AIDS Foundation of South Africa, 2014). Districts with the
highest rates of HIV infection have been shown to be rural ones yet again (The South African Health
News Service, 2012).
TB and HIV/AIDS are very closely connected in that their relationship is often described as a coepidemic (International Federation of Red Cross and Red Crescent Societies, 2015). At least one-third
of the 3.8 million HIV-positive people in the world are also infected with TB (Rodriguez, 2009).
Poverty and poor access to healthcare challenges the successful treatment of those affected by TB. TB
is a highly contagious bacterial infection that is transmitted through the air. Therefore, it is of
importance to adequately isolate or provide for the prevention of spreading this infection. Since TB is
such a contagious disease and accounts for many deaths in rural communities, reducing the spread of
the disease inside the hospital is an important consideration. Additionally, there exist many other
infections that are spread inside hospitals. In Chapter 3 it was found that single rooms in a hospital can
limit the transmission of infections to other patients and it would be wise to have at least enough
single rooms to isolate these patients.
Infant mortality rates in South Africa are six times higher than that of other Organisation for Economic
Co-operation and Development (OECD) countries. The infant mortality rate is 32.6 in urban areas
compared to 52.6 on average in rural areas, whilst in some rural areas the infant mortality is as high as
70.1 (National Department of Health, 2011).
Additionally, rural communities typically have higher rates of mental health disorders too (Fact sheet
rural health, 2013). They are prone to both social and economic disadvantages which are also seen as
high risk factors for depression. High morbidity and mortality rates in poor rural communities also
heighten risk factors for mental disorders such as anxiety and depression. Bad rural mental health is a
handicap to rural development since it places a huge socio-economic burden on poor resource
strapped communities. A leading cause of the neglect of mental health in rural areas is the difficulty
associated with defining these mental disorders (unlike physical conditions) (Kanda, 2013). This is
one of the reasons that the Psychiatry department is essential for district hospitals (refer to R158).
Therefore, it appears that rural areas have specific healthcare needs which differ from urban
communities. For this reason, it is necessary to allow the designer of a hospital to change the hospital
layout and rooms according to the healthcare needs, e.g. if TB is the most common illness in a
community over a prolonged period of time, it may be beneficial to design the hospital in a way that
accommodates many TB-patients.
Attitudes towards health
Attitudes to health and illness generally differ between urban and rural communities, according to
Strong and Elliot-Schmidt (1997). The detrimental nature of illnesses and disabilities in urban
104
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
communities is mainly attributed to discomfort from pain or negativity regarding loss of cosmetic
attractiveness and ability to live a full life. However, for rural communities the response to disabilities
or illnesses is often related to the degree with which the disability or illness impacts their productivity
negatively.
According to Strong and Elliot-Schmidt (1997), rural inhabitants exhibit the perception that if their
health diminishing condition is not life-threatening then visiting the hospital may as well be postponed
until an appointment can be fit into his/her routine (which more often than not delays treatment
indefinitely). These perspectives compound their health-negating behaviours. Rural communities
value independence and self-reliance much more than their urban counterparts (Gessert, Waring,
Bailey-Davis, Conway, Roberts & Van Wormer, 2015). Recently this has been supported by the high
death toll in the Eastern Cape whereby young men during cultural initiation rituals enlisted
unqualified traditional doctors to perform operations on them whereby infections resulted in their
deaths (Bullock, 2015; Makinana, 2015).
In terms of the impact that rural communities attitudes have on hospital floor layouts, it only acts so as
to reinforce the argument for facilities which allow multiple operations to be conducted in a short
period of time. Therefore, the flow rate is again seen as a crucial factor in organising the floor layout.
4.4 Floor layout implications
In the previous section it was found that the main healthcare challenges faced by rural communities
include a lack of funds, high indirect costs of patients, and shortages of healthcare staff. But the
question is one of how this relates to the layout of a hospital.
When one considers the lack of funds available to operate or maintain a hospital, it would make sense
to minimise costs. Therefore, minimising costs is identified as one of the hospital design objectives
specific to rural hospitals. It is important to understand the different types of ‘cost’ involved in a
hospital. Consider the variables used in Chapter 2 to design a hospital layout: 𝛾𝑖𝑗 , 𝑐𝑖𝑗 , and 𝑏𝑖𝑗𝑘𝑙 . The
variable 𝛾𝑖𝑗 refers to the cost of locating and operating department 𝑖 at location 𝑗. The variable 𝑏𝑖𝑗𝑘𝑙 is
the monetary term of locating department 𝑖 at location 𝑗. The variables 𝛾𝑖𝑗 and 𝑏𝑖𝑗𝑘𝑙 are not dependent
on the position of the department within the hospital, and therefore do not need to be optimised when
designing the layout. The variable 𝑐𝑖𝑗 refers to the cost of transporting a patient between departments.
In a factory setting, transportation using conveyor belts or pallet jacks will be relatively easy to
identify and calculate. However, in a hospital setting this is not very straightforward. Patients are often
accompanied when being transported between hospital departments by nurses or doctors which takes
time away from their work, e.g. mechanically ventilated patients should be accompanied by a critical
care nurse as well as a doctor. Therefore, transportation costs can possibly be determined as a
segment of staff salary (Branson & Blakeman, 2013). However, this data is difficult to generate and it is
105
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 4 | RURAL VERSUS URBAN HOSPITALS
only available after building the hospital which is not available when designing a hospital layout.
Another approach is to calculate transportation costs is using the product of distance and flow
between two departments (Bidyarthy, 2012). This appears to be the best explanation for
transportation cost and will be used from this point onwards. The variable 𝑐𝑖𝑗 is thus dependent on the
layout of the hospital.
On the other hand, high indirect costs incurred by patients due to accommodation expenses, long
travel distances, and lost income due to waiting times in inter-hospital (external to the hospital)
expenses, do not influence the sizes of hospital rooms or where each hospital department is located.
Therefore, it cannot be one of the hospital design objectives. Nevertheless, this problem can be
addressed by making public transport available for rural communities.
Shortages of staff as the result of long patient queues and overworked staff suggests that efficiency,
i.e. patient flow, should be optimised. This forms the second hospital design objective specific to rural
hospitals. Additionally, resources can be optimised by identifying the bottlenecks (usually identified as
the doctors) in the system, and reducing them. This however is not in the scope of this study, but it is
recommended to apply the popular industrial engineering concept of Theory of Constraints to remove
such bottlenecks.
Lastly, it is clear that rural communities have different health needs to urban communities and for this
reason it is important that the user of the framework developed in this study, is able to change the
layout accordingly.
4.5 Conclusion
This chapter investigated the differences and similarities between urban and rural hospitals, including
the numerous constraints applicable to the various hospital rooms. The key findings are summarised
in the paragraphs that follow.
The most important standards and regulations associated with hospital design pertaining to the floor
layout were found to be the National Core Standards and Regulations 158. Using these regulations and
others found in literature, the minimum room sizes and departments of a district hospital were
identified.
After this, the specific difficulties faced in rural settings were deliberated upon to arrive at what
appears to be the most important considerations for the layouts of rural hospitals: minimise costs and
patient flow through the hospital. Furthermore, it was found that infection prevention is an important
consideration and single rooms should therefore be incorporated in the framework where possible.
Chapter 5 develops the layout design framework specifically tailored to rural hospitals.
106
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5
PROPOSED LAYOUT DESIGN FRAMEWORK FOR
RURAL HOSPITALS
“Consequently, any discussion (including this one) can only be a partial description of possibilities, but a
review of several major interpretive frameworks can provide a sense of options.”
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
Creswell (2008)
107
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK
FOR RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
5.1 Introduction
In Chapters 2 and 3, quantitative and qualitative methods for designing the layout of facilities and
hospitals were discussed respectively. In this chapter these two methods are integrated. More
specifically, a suitable layout model is chosen and adapted to accurately represent the real world
problem while taking into consideration the needs and specific challenges of rural communities as
discussed in Chapter 4. The assessment of model objectives, assumptions, inputs, and outputs as well
as qualitative design considerations are used as criteria. Consequently, an adequate solution method
to optimise the chosen model is selected. Lastly, the generic rural layout design framework is
developed using Excel VBA, and RStudio.
5.2 Integration of quantitative and qualitative methods
As this study employs a mixed methods strategy, determining whether it is possible to integrate the
quantitative and qualitative methods with one other, and how to do this, is necessary. The quantitative
methods research in this study involves layout models, each exhibiting an objective function, a set of
constraints, inputs, and outputs. The qualitative methods involve a relationship diagram illustrating
the functional interaction of each hospital department. Furthermore, the main hospital design
considerations were identified and linked to quantitative methods (see Section 3.4.6). These
considerations will be incorporated as criteria for selecting the most adequate layout model (see
Section 5.5.5).
A possible way to integrate the quantitative and qualitative methods is to use the results of the
qualitative methods (resulting relationship diagram) as an input for the quantitative methods. This is a
known mixed method phenomenon called embedding (otherwise known as a concurrent embedded
strategy). Embedding has a primary method guiding the project and a secondary database that plays a
supporting role in the procedures (Creswell & Clark, 2009). This typically means that the secondary
method and primary method addresses different questions or seeks information at a different level of
analysis. This type of research is ideal for the researcher to gain broader perspectives as a result of
using different methods. In this study, the primarily qualitative data are used to aid the quantitative
study and to represent the real world problem more accurately. The integration of these two methods
is discussed in more detail in Section 5.5.7.
Since this research develops a conceptual framework, it forms part of concurrent transformative
methodology. The transformative approach is guided by the researcher’s use of a specific theoretical
perspective based on ideologies such as a conceptual framework, critical theory, advocacy, or
participatory research (Creswell, 2015). This perspective is reflected in the research objectives and
aims discussed in Section 1.3. According to Clark and Ivankova (2015) the concurrent transformative
model may take on the design features of an embedded approach. Therefore, this thesis follows a
108
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
concurrent transformative research approach with embedded design features (embedding the data).
This type of research has the advantage of positioning mixed methods research within a
transformative framework (Creswell & Clark, 2009).
5.3 Theoretical framework definition
Theoretical or conceptual frameworks can be compared to maps (Dewey, 2013). Maps are problemsolving tools that help the user navigate through experiences. Maps represent an abstract form of
reality and when accurate, they enable navigation within that reality (Shields & Tajalli, 2006). Dewey
further points out that due to the fact that problems are constantly changing conceptual tools are
required to be constantly refashioned so as to meet new demands.
Jabareen (2009) clearly defines a theoretical framework as “a network, or ‘a plane,’ of interlinked
concepts that together provide a comprehensive understanding of a phenomenon or phenomena. The
concepts that constitute a conceptual framework support one another, articulate their respective
phenomena, and establish a framework-specific philosophy.”
It is worthwhile understanding the difference between models and frameworks. Ilott, Gerrish, Laker
and Bray (2013:1) distinguish between the two by saying that frameworks “are descriptive, showing
relevant concepts and how they relate to each other” whereas models tend to be more “prescriptive,
specific and with a narrow scope.”
Frameworks can be classified into five types (Shields & Tajalli, 2006; Shields, 1998):

Working hypothesis: exploration or exploratory research;

Descriptive categories: description or descriptive research;

Practical ideal type: analysis (gauging);

Models of operations research: decision making; and

Formal hypothesis: explanation and prediction.
The purpose of the framework developed in this study is to guide or help a user (likely the architect or
hospital planner) in making decisions regarding the layout of a rural hospital, e.g. where to place each
department, how big each department should be, what the approximate cost of the design will be, etc.
Therefore, it is clear that what needs to be developed is a decision making framework. To do this,
‘models of operations research’ are used to determine what the best layout for a rural hospital in a
specific context is. The purpose is to guide the user in making decisions and not to replace him/her.
5.4 Real world problem
As discussed in Chapter 1, the real world problem involves designing the optimal layout of a rural
hospital. Decisions need to be made about which departments to include, what size each department
109
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
should be and where each department should be positioned. The proposed rural hospital design
technique will integrate quantitative as well as qualitative design methods, and will be presented in
the form of a block layout.
In Chapter 4, the needs and specific challenges of rural communities were discussed. It was found that
the main problems regarding healthcare faced by these communities include a lack of funds, high
indirect costs to patients, and shortages of healthcare staff. These problems form the objectives of the
real world problem. However, the layout of a hospital cannot influence all of these problems as
discussed in Section 5.4.1.
The real world problem also involves the following decisions:

The floor space used for hospital corridors;

The departments to include;

The location of each department’s centre;

The number of beds in each department;

The number of beds in the hospital;

The rooms to include in each department;

The size of each department;

The size of each room within each department; and

Which departments should and should not be placed in close proximity to each other.
5.5 Selection of layout model
A layout model is proposed that is most suited to represent the real world problem based on layout
objectives, assumptions, inputs, outputs, and design considerations. Values of 1, 0, -1, and -M will be
rewarded for each layout model’s objective, assumption, input, and design consideration according to
how well it represents the real world problem. A value of 1 indicates an accurate, valid representation,
a value of 0 shows a neutral representation, -1 indicates an inaccurate representation, and –M (where
M is a large positive number) shows that the model should not be chosen to represent the real world
problem. In Section 5.5.6 all these awarded values are compared, surmised, and the most adequate
model is selected. Lastly, necessary adjustments are made to this model.
5.5.1 ANALYSIS OF LAYOUT MODEL OBJECTIVES
This section analyses and compares the objectives of a rural hospital layout, and the objectives of the
layout models discussed in Chapter 2.
Real world problem
The purpose of any hospital is clearly to optimise the health outcomes of the surrounding community,
namely to deliver high quality services to as many people as possible. If one assumes that the quality of
services stay constant, the aim should be to maximise the number of patients that go through the
110
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
hospital over a certain period (known as throughput). Throughput is defined by Business
Dictionary (2016) as “the productivity of a machine, procedure, process, or system over a unit period.” In
this case the number of patients that are either cured, referred, treated, or passed away minus the
number of patients that enter the hospital seeking treatment measured over a specific period.
However, it is also important to differentiate between the objective(s) of a hospital as a whole and the
objective(s) of a hospital layout. The hospital layout has a limited influence, e.g. it can influence the
travelling time/distance of patients between departments, but not the operating costs or building
costs of hospital departments (assuming that it is a single floor layout as discussed in Chapter 1).
Building costs stay constant regardless of the position of a department relative to another. The
travelling time/distance can be influenced by each department’s position. Patients will always visit
hospital departments in a certain order regardless of each department’s location, as seen in Figure 19
of Chapter 3. Since the goal of this study is to maximise throughput, it is necessary to maximise the
number of patients that pass through the hospital.
This point was confirmed in Chapter 4 by concluding that the objective of designing the layout of a
district hospital is primarily to optimise flow or cost (product of distance and flow). However, patient
flow is not a variable one can influence directly since it is an input to the layout models. But what can
be controlled is the distance between departments and arrangement in which they are located.
Varying corridor widths to improve flow is hardly ever a solution to improve flow since corridors are
designed according to national regulations. Logically, when deciding upon the objective of a layout,
one needs to link flow (flow rate) to the location of each department since a decision needs to be made
as to where to locate each department. This can be accomplished by optimising the material handling
costs as discussed in Chapter 2.
Layout model objectives
Some popular layout models found in literature for the FLP were discussed in Chapter 2. The main
objective in each of these models is singular, as shown in Table 31. Some researchers have considered
multi-objectives to solve the FLP. For example, Dweiri and Meier (1996) formulated a FLP that
simultaneously minimises the material handling flow with the equipment flow and the information
flow. Other authors combine different objectives into a single one by means of an Analytic Hierarchy
Process methodology (Yang & Kuo, 2003), or by using a linear combination of the different objectives
(Chen & Sha, 2005). The Analytic Hierarchy Process methodology is a very useful aid in decision
making of problems involving multiple criteria. Since the objective of the real world problem is to
optimise material handling cost, a single objective model will be chosen and multi-objective models
will not be investigated.
111
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
TABLE 31: ANALYSIS OF LAYOUT MODELS' OBJECTIVES
Layout model
Quadratic Assignment
Problem
Quadratic Set Covering
Problem
Linear Integer
Programming Problem
Mixed Integer
Programming Problem
Graph Theoretic Problem
Linear Continuous Model
Linear Mixed Integer Model
Model objective
Variable(s)
related to flow
Objective function
F
Minimise total cost
𝑓𝑖𝑘 , 𝑐𝑖𝑗
Flow x transportation cost
1
Minimise fixed cost and/or
flow
𝑓𝑖𝑘 , 𝑑𝑗𝑙
Flow x distance
1
Minimise total cost
𝑓𝑖𝑘
Flow x transportation cost
1
Minimise total cost
𝑓𝑖𝑘 , 𝑐𝑖𝑗
Flow x transportation cost
1
-
Closeness ratings
1
Maximise a closeness rating
measure
Minimise transportation
cost
Minimise total traveling time
𝛼𝑖𝑗 , 𝑐𝑖𝑗
𝛼𝑖𝑗 , 𝑡𝑖𝑗
Transportation cost x
frequency x distance
Time x frequency x
distance
1
1
Six of the seven layout models as shown in Table 31 aim to optimise material handling costs which
correspond to the objective of the real world problem. The GTP aims to optimise a closeness rating
which is based on the desirable adjacency of each pair of facilities. This can theoretically also be based
on material handling cost and is therefore also deemed an adequate model according to its objective.
The objective function of the QSCP consists of two terms that do not have the same units. The
parameter theta is incorporated which may be varied in order to test different conversion rules.
5.5.2 ANALYSIS OF LAYOUT MODEL ASSUMPTIONS
The following assumptions regarding the real world problem were made (as discussed in Section 1.4):

The hospital will be designed for a single floor layout; and

The optimal positioning of the hospital departments will be determined and not the positioning of the
rooms inside of each department.
The assumptions of the seven layout models were given in Chapter 2. Table 32 summarises this.
112
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
TABLE 32: ANALYSIS OF LAYOUT MODELS' ASSUMPTIONS
Equal sized departments
The departmental flow is independent of the
departments’ locations
The cost to transport one patient is independent of
the departments(2)
The desirability of locating each pair of departments
adjacent to each other is known
x
x
-1
x
x
0
x
x
-M
x
x
x
x
x
x
x
x
x
x
x
x(1)
1
x
-1
x(2)
The shape of the departments is known in advance
The departments are square or rectangular in shape
x
There is no restriction on the shape of the building in
which the facilities are to be located
x
1-M
Total
x
1
x
Area and shape does not influence solution
The departments are oriented in only two given
directions
Value
x
Linear Mixed Integer
Model
x
Linear Continuous
Model
x
Graph Theoretic
Problem
x
Mixed Integer Problem
Linear Integer
Programming Problem
The available space can be divided into blocks of
equal size
The candidate locations for each centre have to be
established in advance (discrete layout)
The number of departments is equal to the number
of locations
Quadratic Set Covering
Problem
Assumption
Quadratic Assignment
Problem
Layout models
x
x
1
x
x
-1
x
x
-1
1
x
x
x
x
x(3)
x
x
x
2
1-M
2
1
1
1
(1) Uses closeness graph which is independent of locations.
(2) Constraints can be added to ensure this.
(3) Square/rectangular shapes in the horizontal and vertical axis.
As mentioned in Chapter 2, the layout models can be classified as either discrete or continuous layouts
based on the nature of centre locations of the departments. Discrete layout models limit the
departments to be located at a number of discrete points while continuous layout models allow new
departments to be located anywhere within the modelled space. The discrete layout models under
consideration (QAP, QSCP, LIPP) divide the available space into smaller blocks. Thus, all of these
models assume that the centres of departments can only be placed at the nodes of the network.
Therefore a value of -1 is assigned to these layout models. This assumption is similar to the next one
which assumes the candidate locations for each centre are established in advance (therefore a value of
0 is assigned). The MIPP, GTM, LCM, and LMIM allow departments to be placed anywhere in the
modelled space.
113
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
In Chapter 4, the minimum area required for each department was determined. These areas differ to a
great extent from one another. Thus, assuming equal-sized departments for a hospital is not a realistic
assumption. Consequently, a value of –M is assigned to the QAP and LIPP (M being a very large positive
number) in order to ensure the model is not chosen. The QSCP on the other hand is able to
accommodate different centre sizes. Several candidate locations are made available for the location of
each centre.
The assumptions of independent flow and transportation costs of the department’s locations are valid
assumptions since patients cannot decide which department they want to visit or in what order. They
are sent to each department via administrative staff, doctors, or nurses. Therefore, patients will visit
certain departments regardless of the department locations. The only choice patients have is through
which door they will enter, namely the main entrance, emergency entrance, or outpatient entrance as
seen in Figure 12 (Chapter 3). This however is not between departments and does not contradict the
assumptions. These two assumptions are similar and a value of 1 is awarded to one or both of these
assumptions. The Graph Theoretic Problem (GTP) does not use flow or cost variables, but it uses a
closeness measure graph which can be defined as the departmental flow.
The assumption that departments are oriented in only two given directions is valid since many
hospitals are only built in two directions. It is rare that any building is built in three or more directions.
This simplifies the building process, furnishing of rooms, and flexibility of rooms. Assuming square or
rectangular shaped departments imposes a restriction on the layout solution. In reality many
departments have near square or rectangular shapes, L-shapes, or even U-shapes.
According to the layout model assumptions, it is clear that the QSCP, and Mixed Integer Problem are
more suited to model the real world problem.
5.5.3 ANALYSIS OF LAYOUT MODEL INPUTS
The inputs used in the layout models were discussed in Chapter 2 and are summarised in Table 33.
114
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
TABLE 33: ANALYSIS OF LAYOUT MODEL’S INPUTS
locating
and
x
0
x
x
x
x
1
x
x
The flow of patients between departments
The cost of transporting patients between
departments
Distance between the centroids of candidate
locations
Number of blocks into which the total area occupied
by all departments is divided into
x
x
1
x
Set of candidate locations for departments
1
x
x
x
x
x
x
-1
The closeness rating indicating the desirability of
locating departments adjacent
The frequency of trips to be made between
departments
x
The length and width of each department
The minimum distance by which departments are to
be separated
The time required for a patient to move between
departments
1
x
x
1
x
x
-1
x
1
x
Total
1
1
Value
x
Linear Mixed Integer
Model
x
Linear Continuous
Model
x
operating
Graph Theoretic
Problem
Mixed Integer Problem
of
Linear Integer
Programming Problem
The fixed cost
departments
Quadratic Set Covering
Problem
Inputs
Quadratic Assignment
Problem
Layout models
1
2
1
2
1
1
The input of ‘the fixed cost of locating and operating departments’ is constant regardless of where a
department is positioned. Therefore, a value of zero is assigned to this input. The flow of patients
between departments is an essential input since it is the objective of the selected layout model. The
inputs of distance between the centroids of candidate locations, and the cost of transporting patients
between departments are significant since they correspond to the flow assumption. The number of
blocks into which the total area occupied by all departments is divided affects the complexity of the
problem as well as the accuracy of the solution. Candidate locations for departments, i.e. discrete
points, restrict the solution space and are undesirable and therefore assigned -1. The closeness rating
indicating the desirability of locating departments adjacent to one another is significant since it
corresponds to the flow assumption. The frequency of trips to be made between departments is
important and corresponds to the flow assumption. The length and width of each department restricts
the solution space. The time required for a patient to move between departments is important and
corresponds to the flow assumption.
5.5.4 ANALYSIS OF LAYOUT MODEL OUTPUTS
Table 34 shows an analysis of the seven layout models’ outputs.
115
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
TABLE 34: ANALYSIS OF LAYOUT MODELS' OUTPUTS
Quadratic Set Covering
Problem
Linear Integer
Programming Problem
Mixed Integer Problem
Graph Theoretic
Problem
Linear Continuous
Model
Linear Mixed Integer
Model
Block layout
x
x
x
x
x
x
x
1
Location of each department
x
x
x
x
x
x
x
1
Minimum total cost
x
x[1]
x
x
Outputs
Minimum closeness rating
x[2]
1
x
1
Minimum total time
x
Minimum flow between departments
x
3
3
1
1
May have empty spaces between departments
Total
Value
Quadratic Assignment
Problem
Layout models
3
3
3
x
x
2
2
-1
(1) Refers to fixed costs only.
(2) Refers to transportation costs only.
The Linear Mixed Integer model (LMIM) and the Linear Continuous Model (LCM) do not consider the
shape and dimensions of the building and may produce a layout which contains empty spaces between
the departments. This is not ideal and a negative value is awarded. The rest of the outputs accurately
produce a layout and as such a value of 1 is awarded to each. Take note that the fixed cost of locating
the department does not influence the position of the department within the layout, thus a value of 0 is
assigned to it. A value of 1 is assigned to transportation costs.
5.5.5 ANALYSIS OF DESIGN CONSIDERATIONS
The qualitative methods in Chapter 3 were compared with the layout models. Table 35 summarises
the design measures that each layout model addresses.
116
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
TABLE 35: ANALYSIS OF LAYOUT MODELS' DESIGN CONSIDERATIONS
x
x
x
x
x
x
Efficiency
Minimise cost
x
x
x
x
x
Flexibility and
expandability
Sustainability
x
x
x
5
x
5
Minimise travel distances
Minimise costs
Total
x
5
Value
Mixed Integer
Programming Problem
x
x
Linear Mixed Integer
Model
Linear Integer
Programming Problem
x
x
x
Design measures
Linear Continuous
Model
Quadratic Set Covering
Problem
Optimise occupant flows
Minimise transportation
Minimise waiting times
Design
objectives
Graph Theoretic
Problem
Quadratic Assignment
Problem
Layout model
x
x
x
1
1
1
1
x
x
4
x
x
5
0
1
1
4
5.5.6 SELECTION OF LAYOUT MODEL
From Table 36 it is evident that the QSCP and the Mixed Integer Problem are both equally the most
suitable models to represent the real world problem.
TABLE 36: SELECTION OF THE BEST LAYOUT MODEL
Quadratic Assignment
Problem
Quadratic Set Covering
Problem
Linear Integer
Programming Problem
Mixed Integer Problem
Graph Theoretic
Problem
Linear Continuous
Model
Linear Mixed Integer
Model
Layout models
1
1
1
1
1
1
1
1-M
2
1-M
2
1
1
1
Layout model inputs
1
1
1
2
1
2
1
Layout model outputs
3
3
3
3
2
2
3
Design considerations
5
5
5
4
0
5
4
11-M
12
11-M
12
5
11
10
Criteria
Layout model objectives
Layout model assumptions
Total
An advantage of using the Mixed Integer Problem lies in that it is a continuous layout as opposed to a
discreet layout used in the QSCP. This implies that the departments can be located at any point in the
solution space. However, the solution space is considerably larger than in the case of a discreet layout
making solving the model significantly harder. Furthermore, it was found in Chapter 3 that a minimum
of 16 and a maximum of 19 departments are required for a rural hospital which makes it a very large
problem to solve. Thus choosing a smaller solution space (QSCP) is beneficial. Furthermore, this study
117
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
designs a block layout as opposed to a detailed layout and therefore the QSCP solution is sufficient. The
QSCP also allows L-shape and U-shape departments in the final solution. For these reasons the QSCP is
chosen to represent the real world problem.
5.5.7 ADJUSTMENTS TO LAYOUT MODEL
In Chapter 3 a relationship chart based on the activity requirements of each department was arrived
at. The chosen layout model, the QSCP, uses patient flow as an input for the model and as a means of
determining which departments to put relatively close and far from each other. If this is the case,
departments which are seldom or never used by patients will most likely be placed on the outskirts of
the layout, e.g. laundry. However, it makes logical sense to place the laundry department close to the
wards since bed sheets need to be replaced every time an inpatient leaves the hospital. Clearly, patient
flow is not the only important consideration, as flow in a hospital also includes medical staff flow,
cleaning staff flow, specimen flow, test results flow, and waste flow.
Another way to explain this is to realise that some patients are in a more critical condition than others,
e.g. requiring emergency treatment. There might not be many critical patients that are transported
between two departments and the flow therefore is low, but it is essential to make the distance
between such departments as short as possible. Consequently, using patient flow as a means to
determine the relationships between hospital departments is not necessarily an accurate
representation of the real world problem.
A possible solution is to use the relationship chart arrived at in Chapter 3 which is based on each
department’s activities as well as logical reasoning to replace the flow aspect of the model.
5.6 Selection of solution method
The solution method depends on the problem size, i.e. the number of departments. In Chapter 4 the
number of required departments was determined as being 16, with two being optional. Two types of
solution methods were identified, namely exact methods, and metaheuristics (discussed in
Section 2.5). Branch and Bound Algorithm (an exact method) can find optimal solutions only up to a
problem size of 16 while metaheuristics approximate the optimal solution for larger problems. Since
the problem size ranges from 16 to 19, the Branch and Bound Algorithm will be used to solve smaller
problems and a metaheuristic will be used to solve larger problems.
Metaheuristics can be classified into P-metaheuristics and S-metaheuristics (refer to Table 21). Pmetaheuristics can improve a population of solutions (candidates) while S-metaheuristics seek to
improve a single solution. Generating one solution is much quicker and easier than generating a
population of solutions. For this reason an S-metaheuristic will be chosen.
118
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
The LS method is quick and greedy, but not accurate for solving problems with numerous locally
optimum solutions. Therefore Simulated Annealing (SA) and Tabu Search (TS) are possible solution
methods. TS generates a tabu list which complicates the solving process more than SA. For this reason
SA is chosen.
5.7 Development of layout design framework
In order to guide the user to design a near-optimal hospital layout, two computer programs have to be
developed. The first program should be able to capture input from a user and calculate the sizes of
each hospital department accordingly, while the second is required to solve the layout using these
inputs. In this section a few assumptions are outlined, followed by a discussion on the development of
the framework via eight steps. In conclusion, a few examples of the outputs of this framework are then
discussed.
5.7.1 ASSUMPTIONS
When developing a framework there are invariably assumptions which have to be made, and these
include that:
1.
The framework is developed for a single floor layout (see Section 1.7);
2.
A district hospital layout is to be developed which implies a specific set of required and optional
departments and rooms (see Section 1.7); and
3.
The shape of each hospital department is independent of the arrangement of the rooms inside it which
implies that the dimensions of each department can change without the need to analyse the
arrangement of rooms within the department (see Section 1.7).
5.7.2 SELECTION OF PROGRAMS
Herein the two programs required to develop the layout design framework will be described. As
mentioned, the first is required for capturing inputs from the user to calculate the sizes of each
department, and the second is required to take these inputs and utilise them to optimise the
arrangement of departments, thus creating as many iterations of the floor layout as are required.
The user input program must be capable of making various calculations and suggesting ballpark
values to the user of the framework (e.g. number of beds, sizes of rooms). The user should then be able
to change these values according to preferences and case-specific insights. These values should then
be captured in the program and the total size and building cost of each hospital department should be
calculated. Excel VBA was chosen for these purposes since it is a convenient software program for
developing user-friendly interfaces and making the required calculations.
The layout construction program seeks the optimal (or near optimal arrangement) of the hospital
departments specified in the first program. The QSCP (see Section 5.5) was chosen as the most
119
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
adequate layout model for the framework, utilising the Branch and Bound Algorithm and SA as the
solution methods (see Section 5.6). The Branch and Bound can optimally solve the QSCP with up to 16
departments and SA can approximate the optimal solution for larger problems. Thus, the second
program has to solve the QSCP with SA or the Branch and Bound Algorithm while using the inputs
generated in the first program. Lingo was selected as the programming language for the Branch and
Bound Algorithm while RStudio was selected for programming SA.
5.7.3 FRAMEWORK STEPS
The development of the framework is discussed via eight steps, outlined in Figure 21, which indicates
the calculations made in each of the programs (i.e. Excel VBA, Lingo, and RStudio). The first six steps
are executed in Excel VBA. Each step (and sub-step) represents a window that is prompted to the user
of the program. Each window recommends values to the user of the program and allows them to be
changed by the user. The blue text in Figure 21 represents all the variables that can be edited by the
user. The eight steps of the framework will now be discussed.
120
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
1. Inputs
Rural population size
Occupancy rate
Circulation space
Construction cost
Optional departments
5. Hospital departments
5.1 Administration
5.2 Obstetrics
5.3 Operating Suite
5.4 Paediatric Unit
Suggest total number of
beds
5.5 Outpatient Unit
5.6 Laundry
Required
2. Hospital beds
Number of hospital beds
Suggest number of beds per
department
5.7 Kitchen
5.8 Sterilisation & Disinfection Unit
5.9 Pharmacy
5.10 Emergency and Casualty Unit
5.11 Acute Psychiatric Facility
3. Beds per department
5.12 Chronic Care Unit
Number of beds per
department
5.13 Rehabilitation Unit
5.14 Mortuary
Suggest relationships
between departments
5.15 Laboratory
4. Relationship chart
5.17 Neonatal Intensive Care Unit
Optional
5.16. Radiology
Departmental
relationships
5.18 Intensive Care Unit
5.19 High Care Unit
Suggest number of beds
Suggest minimum room areas
Suggest number of single rooms
Yes
6. Summary
Area per department
Cost per department
Calculate total hospital area
Calculate total cost
Change inputs
No
EXCEL VBA
Yes
7. Branch and Bound
Calculate optimal layout
using Branch and Bound
method
Number of beds
Number of single rooms
Optional rooms
Size of each room
Calculate total area per
department
Calculate departmental
construction cost
No
< 17 departments
LINGO
RSTUDIO
8. Simulated Annealing
Calculate optimal layout
using Branch and Bound
method
FIGURE 21: FRAMEWORK OUTLINE
Step 1 Inputs
The purpose of the first window is to define the characteristics of the hospital layout. In this window
the user is prompted to enter the rural population size (this is the only variable for which a baseline
value is not suggested to the user). The required hospital departments are shown and the user can
select the optional departments by clicking on the checkboxes next to each listed department. The
window also recommends the following variables that can be changed by the user: occupancy rate,
circulation space, and construction cost. These variables will now be discussed further.
121
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
i.
Occupancy rate
The occupancy rate is defined by Segen's Medical Dictionary (2011) as “the proportion of beds in a
hospital or ward which are in use at a particular time.” The Organisation for Economic Cooperation and
Development (2005) (OECD) considers a rate of 85% as the limit of safe occupancy. Bagust, Place and
Posnett, (1999) modelled the dynamics of the hospital system, using a discrete-event stochastic
simulation model which reflects the relation between demand and available bed capacity and found
that increased risks are apparent when average bed occupancy rates exceed approximately 85%, and
furthermore that “an acute hospital can expect regular bed shortages and periodic bed crises if average
bed occupancy rises to 90% or more.” An acute hospital is defined by Hirshon, Risko, Calvello, De
Ramirez, Narayan, Theodosis, and O’Neill (2013) as “a short-term hospital that has facilities, medical
staff and all necessary personnel to provide diagnosis, care and treatment of a wide range of acute
conditions, including injuries.” This 85% figure dates back to nearly 100 years ago when Erlang
developed the queuing theory which argues that systems are most efficient when they operate at 85%
capacity (Bain, Taylor, McDonnell & Georgiou, 2010). Another method to optimising the number of
beds in a hospital is discussed by Goronescu, McClean and Millard (2002) who found that it depends
on the relative value placed upon empty beds and blocked patients. It is therefore prudent to suggest
an occupancy rate of 85% to the user of the framework, but also to allow the user to change this rate
according to personal preferences depending on their hospital requirements (e.g. whether it is likely
to provide more acute or more chronic care, and the cost of empty beds and inadequate space for
patients).
ii.
Circulation space
The importance of circulation space is often overlooked in the designing phase of buildings. Circulation
space is defined by the National Centre for Education Statistics (2006) as “the sum of all areas on all
floors of a building required for physical access to some subdivision of space, whether physically bounded
by partitions or not.” Examples of these spaces include corridors, required exits, traffic aisles,
escalators, stairways, elevator lobbies. These spaces are usually not indicated in block layout plans, yet
careful planning of these areas is essential for an efficient layout (see Section 3.5.2 for the minimum
dimensions required by the law). One way to plan for the necessary circulation space of a layout is to
calculate it as the percentage of the total space. According to Karlen and Fleming (2016) 25% is
recommended for general use of a building.
The United States’ General Services Administration (2012) recommends a method to estimate and
plan for circulation. The benefit of using this method is that the multiplier used in their calculations is
dependent on the proportion of open to enclosed spaces in the layout which makes the estimate more
accurate. They define the following terms:
122
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS

Net Area (NA): the area of each identified program space (i.e. department), including workspaces,
dedicated support (e.g. supply rooms), shared support (e.g. staff break rooms), and special missioncritical support spaces (e.g. laboratories);

Circulation Area (CA): primary circulation is the main circulation route connecting the building core
and common spaces whilst secondary circulation relates to the aisles between individual and support
spaces (e.g. offices, cubicles);

Usable Area (UA): includes NA and CA, but excludes building common spaces such as elevators, exit
stairs, and core toilets. Since this study focuses on a single floor layout, there is no need for elevators or
stairs, and toilets are located within departments. Therefore, the UA is seen as being equal to the total
area required for the hospital layout;

Circulation Multiplier (CM): is applied to the NA to estimate the multiple of CA that should be included
in the UA; and

Circulation Factor (CF): the percentage of UA that makes up the CA.
The CA is a function of the open and enclosed spaces that exist in the layout. Typically, a floor plan
with many open workstations will have a higher CA than a plan made up primarily of enclosed spaces.
Open workstations require a larger area since they require proportionally greater circulation than
enclosed spaces. Therefore, a unique CM should be chosen for each layout.
Take note of the difference between CA and CM: CM is applied to NA while CF is the percentage of US
that makes up the CA. For example, a CM of 1.35 and a NA of 28,000m2 results in a CA of 25.9% of the
total UA. The CM should be selected based on the anticipated amount of CA that will needed for
efficiency. The CM range is 1.4 to 1.6 and the average is 1.5, whereas the CF is between 28% and 38%
with an average of 33%.
From Chapter 4, it is apparent that hospital departments contain mostly enclosed areas for the
purpose of patient privacy. The United States’ General Services Administration (2012) recommends a
CM of 1.4 for a building with a greater portion of enclosed offices and support spaces. Thus, 1.4 (CF of
28.6%) will be used in the framework to calculate the total area needed for each hospital department.
iii.
Construction cost
The construction cost of a hospital layout can be estimated by using the cost per square meter
approach. The IUSS estimated this value for a single story Level 1 (district) hospital to be R14,250/m2.
This value was derived from research conducted on selected case studies where the cost of the
hospital was divided by the construction area. All the case studies were escalated or de-escalated in
order to provide the same base date and comparison of rates. This value excludes value-added tax, but
includes the following costs:

Theatre lights and examination lights;

All fixed systems (e.g. nurse call, fire alarms);

Built in shelves, counters, and nurses stations;
123
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS

All security systems (e.g. closed-circuit television);

All human resource systems (e.g. clocking systems);

IT and communications equipment (e.g. telephones);

Kitchen equipment used in the main kitchen (e.g. fridges);

Laundry equipment used in the main laundry (e.g. washers and dryers); and

Pendants with gas and power connections in theatre and critical care areas.
Another estimation of the building cost of a district hospital in South Africa was proposed by the Africa
Property and Construction Handbook (2013) which lists a cost of $1,110/m2 (exchange rate of R13.45
for $1 on 3 September 2015). This is equivalent to R14,521/m2, which is close to the IUSS
recommendation. However, the route to determining this value is unclear and due to a fluctuating
inflation rate it is a bit uncertain. Therefore, the value recommended by IUSS is used in this study.
Once all the values are entered into the provided spaces and the optional departments are selected (if
desired) in the input window, the user can click the ‘next’ button and continue to the second step.
Step 2 Hospital beds
The purpose of the second window is to suggest the total number of hospital beds to the user and
allow the user to change this value if desired.
Deciding upon the number of beds for a hospital is not a straightforward task. On the one hand there
has to be enough beds available for patients so that under normal operating conditions no patients
have to be refused treatment, and on the other hand empty beds equate to unnecessary costs. In
addition, there is the concern of large variability in the number of occupied beds leading to higher
costs. In order to account for this, some researchers focus on the arrival process of patients at a
hospital and the length-of-stay distribution of patients (Kuijstermans, 2014). As this is not easily
deduced for a rural hospital where these variables are extremely context dependent, another route
should be followed.
Phillip, Mullner and Andes (1984) express the occupancy rate as a function of the hospital size,
percentage of non-urgent beds, relative variation in the demand for hospital care faced by individual
hospitals, and the number of hospitals serving an area. The number of beds serving a specific area
would typically play a significant role when deciding upon the number of beds for an urban hospital,
while this factor is inapplicable to rural communities due to the remote nature of rural areas.
Another way to decide upon the number of beds is to use the average number of beds per capita for
the relevant country. The OECD (2013) provides these statistics of more than 36 countries. According
to Econex (2010), South Africa has an average of 2.5 hospitals beds per 1000 population (including
both public and private sectors). Some of these countries have an excessive hospital capacity while
others are overcrowded. The number of beds required in a country depends on factors such as
124
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
patterns of disease and the availability of alternative care settings (Green, 2002). The international
Inter-Agency Standing Committee (IASC) (2010) gives a vague benchmark for the number of hospital
beds per 1000 population, arguing that it should be greater than one. According to WHO (2010), the
global average is 2.7 beds per 1000 population, where lower and middle income countries have 1.8
and 2.9 hospital beds per 1000 population respectively. They arbitrarily choose a benchmark of 3
hospital beds per 1000 population. However, these values usually include all types of hospitals which
could be misleading. In 2014 the National Department of Health (2014) found that district hospitals in
South Africa have on average 0.7 beds per 1000 population.
Thus, for the purposes of designing a new hospital layout, it is advantageous to express the number of
hospital beds as a function of the community’s population. In order to maintain the balance between
overcrowding and vacant beds, the desired hospital occupancy rate should also be included in this
calculation. The calculation used in developing the layout design framework will now be discussed by
means of an example.
The average length of stay in South African public hospitals is 5.3 days (Ramjee, 2013) and the average
inpatient admissions per 1000 population per year is 58.3 (Harris, Goudge, Ataguba, McIntyre,
Nxumalo, Jikwana & Chersich, 2011). If the community’s population is 200,000, then the average
inpatient admissions per year are 11,660. The average admissions per day are thus 31.95. With an
average length of stay being 5.3 days, 170 beds are needed if the desired occupancy rate is 100%. If the
desired occupancy rate is 85% for the same number of admissions, 200 beds are required. This
equates to 1 bed per 1000 population which falls within the guidelines provided by IASC. This is also
slightly higher than the average number of hospital beds in South African district hospitals.
In conclusion to this step, a number of hospital beds are suggested to the user of the framework in the
hospital beds window according to the specified population size, but the user can still decide to change
this value. Once this is decided, the user can click ‘next’ and proceed to the third step.
Step 3 Beds per department
The purpose of the third window is to suggest the number of beds per department and allow the user
to change these values if so desired.
The number of hospital beds in each department is another important design consideration. A few
references suggest the percentage of beds in a few hospital departments, but no complete guide for
estimating the beds distribution in a hospital was found. A way to estimate missing values is to use the
process of benchmarking. Benchmarking is defined as “a standard or point of reference against which
things may be compared” (Stevenson, 2015). This concept is used to estimate the distribution of beds
in a new hospital using data from a few existing hospitals. Table 37 shows the number of beds of five
South African hospitals and a few references that recommends the number of beds in departments.
125
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
The values that are not suggested by references are estimated using the average of the benchmark
hospitals.
TABLE 37: BENCHMARKING HOSPITAL BED DISTRIBUTION ADAPTED FROM LIFE HEALTHCARE (2016)
Benchmark hospitals
Departments
Obstetrics
Life Carstenhof Hospital
Life Kingsbury Hospital
Life Eugene Marais Hospital
Life Fourways Hospital
Life Glynnwood Hospital
Directorate General of
Health Services (2012)
Directorate General of
Health Services (2012)
Van Reenen (2014)
Kunders (2004)
De Jager (2014)
Percentage of total beds
Other references
21
28
12
16
30
58
78
20%
-
-
20%
Operating Suite
6
12
14
10
10
-
-
-
2%
-
2%
Paediatric Unit
-
-
-
20
-
10
20
-
-
20%
20%
Emergency and Casualty Unit
-
-
-
-
-
10
10
-
-
-
5%
Acute Psychiatric Facility
-
-
-
-
-
10
10
-
-
-
10%
Chronic Care Unit
-
-
-
-
-
66
125
-
-
-
22%*
Rehabilitation Unit
-
-
-
-
80
5
10
-
-
-
4%
Neonatal Intensive Care Unit
12
12
6
5
5
-
-
-
-
-
4%
Intensive Care Unit
10
14
27
26
16
-
-
-
-
-
8%
High Care Unit
7
5
26
8
18
-
-
-
-
-
5%
153
227
364
194
323
150
250
-
-
-
100%
Total number of beds
(1) Adjusted to fill 100%.
The percentages of beds in each hospital department are used to calculate the number of beds in each
department based on the total number of beds that the user selected in Step 2. These calculated beds
per department are then suggested to the user. Once the user decides upon the number of beds in each
department, the ‘calculate’ button will capture these values and use it in the next few steps. If the user
does not specify a value, the suggested values will be used. If the user wishes to return to the previous
step, the ‘back’ button can be clicked. Otherwise, the user can click ‘next’ and proceed to Step 4.
Step 4 Relationship chart
This fourth step specifies which departments should be placed together and which ones should be
separated. Firstly, a matrix of the relationship between each hospital department (generated in
Section 3.4 and discussed in Section 5.5.7) is suggested to the user. The user is able to change each of
these values. This step plays an important role in finding the best arrangement of hospital
departments. After amending the relationship matrix to meet their needs, the user can click ‘next’ and
proceed to Step 5.
126
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
Step 5 Hospital departments
The purpose of Step 5 is to determine the size of each room, allow the user to choose which optional
rooms to include in the layout, determine the total size of each department, and calculate the
construction cost for each department.
In this step a window is shown for each hospital department and optional department (if selected by
the user in Step 1). Each window shows a list of rooms that are required and optional for each hospital
department. The user is able to then select which optional rooms to include in each department by
clicking on a checkbox next to the relevant room. The user also has the option to include a few of the
same rooms in each department (where applicable). The minimum floor space for each room as
required by the South African law is also displayed. The user is able to enter a new area for each room
if desired. Once this is done, the user can click on ‘calculate’ and the program will determine and show
the total area and construction cost of the relevant department. If the user did not specify the area of a
room or if the entered area is smaller than the minimum area, the area provided by law will be chosen
to include in the layout. This will ensure that the layout conforms to the requirements of the law.
Furthermore, each box shows the number of beds chosen for the department and recommends the
percentage of single rooms. The user is also able to change both of these variables. The following boxes
are prompted to the user of the program (the last three optional departments will only be shown if
selected by the user in Step 1):

Administration;

Obstetrics;

Operating Suite;

Paediatric Unit;

Outpatient Unit;

Laundry;

Kitchen;

Sterilisation and Disinfection Unit;

Pharmacy;

Emergency and Casualty Unit;

Acute Psychiatric Facility;

Chronic Care Unit;

Rehabilitation Unit;

Mortuary;

Laboratory;

Radiology;

Neonatal Intensive Care Unit;

Intensive Care Unit; and

High Care Unit.
127
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
The estimation of the size of the Laundry, Kitchen, and Sterilisation and Disinfection Unit work a bit
differently than the other departments. The reason for this is that these three departments do not
contain many rooms and consequently the sum of the minimum areas of these rooms will not give an
accurate estimation for the total size of the department. Therefore, the concept of benchmarking was
used. The calculations used in each of these three departments will now be discussed.
TABLE 38: NUMBER OF BEDS AND APPROXIMATE SIZE OF LAUNDRY ADAPTED FROM FOURIE, SHEARED AND STEYN (2014)
Number of hospital beds
Laundry size (in m2)
200
448
300
570
400
700
500
820
600
945
The main elements in hospital laundries are similar, but the main difference is one of scale. The
Council for Scientific and Industrial Research (1995) (CSIR) compiled Table 38 which shows estimates
for the size of a laundry according to the number of hospital beds. This data was derived from a
number of existing hospitals in the Western Cape.
In order to incorporate the data from the CSIR, the linear graph shown in (67) was calculated to
estimate the size per m2 and this was then programmed into the Laundry box.
(67)
𝑦 = 1.24(𝑏𝑒𝑑𝑠) + 203
The size of a hospital’s kitchen is unique to that hospital, and depends on a variety of factors that
include the number of patients and staff to receive meals, type of patient (e.g. age and culture), type of
menu, and type of food preparation system (Steyn & Boltman, 2014). Table 39 shows the number of
beds in eight South African hospitals along with the kitchen area of each hospital.
TABLE 39: NUMBER OF BEDS AND SIZE OF KITCHEN
Hospital
Number of beds
Kitchen size (in m2)
Moses Kotane
183
292
Holy Cross
231
588
George
256
758
Worcester
269
276
Khayelitsha
277
625
Bertha Gxoba
300
565
Paarl
365
581
Natalspruit
752
1387
The data shown in Table 39 was used to generate a linear graph for the estimated size per m2 shown
in (68).
128
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
(68)
𝑦 = −0.0003(𝑏𝑒𝑑𝑠) + 1.9863
Similar to a hospital’s kitchen, the Sterilisation and Disinfection Unit is unique to the requirements of
that hospital. Space determinants of this unit include the number of theatres, surgical procedures,
distribution system, and processing needs (Steyn & Sheard, 2004). Table 40 shows data recommended
by the Department of Health (2001).
TABLE 40: NUMBER OF BEDS AND SIZE OF STERILISATION AND DISINFECTION UNIT ADAPTED FROM STEYN AND SHEARD (2004)
200
Sterilisation and Disinfection Unit size
(in m2/bed)
0.7
400
0.6
600
0.5
800
0.5
1000
0.5
1200
0.5
Bed size
Since the data from Table 40 does not form a linear graph, it is decided to use a value of 0.7m2/bed for
estimating the size of the Sterilisation and Disinfection Unit of the framework as a district hospital
typically has between 0 and 200 beds.
Once all boxes of the hospital departments are completed, the user can continue to the last step in
Excel VBA.
Step 6 Summary
In this step a window is presented to the user with the total size of each hospital department as
determined in Steps 1 to 5. The area of each department with the circulation space included is also
shown. Using these values, a new construction cost per department is calculated. The total
construction cost and size of the hospital is also shown. The user is also able to return to Step 1 and
change values if so desired. The old entered values will still be shown so that the user does not have to
redo the whole process.
129
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
TABLE 41: EXAMPLE OF EXCEL VBA OUTPUT
Sum of room sizes
(in m2)
148
Circulation space
included (in m2)
206
Obstetrics
435
604
Operating Suite
118
164
Paediatric Unit
230
319
Outpatient Unit
372
517
Laundry
148
206
Kitchen
167
232
Sterilisation and Disinfection Unit
106
147
Pharmacy
120
167
Emergency and Casualty Unit
439
610
Acute Psychiatric Facility
174
242
Chronic Care Unit
520
722
Rehabilitation Unit
233
324
Mortuary
76
106
Laboratory
88
122
Radiology
145
201
Neonatal Intensive Care Unit
48
67
Intensive Care Unit
91
126
High Care Unit
90
125
3 748
-
Total size of hospital
-
5 206
Total construction cost
-
75 693 983
Departments
Administration
Total area of hospital rooms
This step concludes the code in Excel VBA. An example of the output of the Excel VBA program is
shown in Table 41. A circulation space of 28% was included in the layout.
In order to explain how the framework operates, four examples will be discussed. The output of Steps
1 to 6 will be used as input for Steps 7 and 8. These two steps use the QSCP formulation with the
Branch and Bound Algorithm and SA respectively.
A hospital size of 5,206m2 can be accommodated by a site space of 75 by 75 meters (5,625m2). It was
decided to use these dimensions and divide the area into 3 by 3 meter blocks. This means that there
are 576 potential locations where the centre of each hospital department can be placed. Consequently,
the distance variable 𝑑𝑗𝑙 will be a matrix with 331,776 (576 x 576) entries. The relationship chart from
Step 4 is used as an input 𝑓𝑖𝑘 to the framework. The layout problem will now be solved.
Step 7 Calculate the optimal layout using Branch and Bound Algorithm
Lingo is a convenient program for using the Branch and Bound Algorithm, and was therefore chosen.
The Branch and Bound Algorithm is able to solve the FLP optimally, but only up to a problem size of
16. The QSCP was first coded in Lingo for a problem size of 6, but the program could not handle the
130
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
amount of variables and constraints. The program has a limit of 150 constraints and 300 variables
(373 constraints were found and 576 variables). The improved version of this program will also not be
able to handle this problem having 250 constraints and 500 variables. Since Lingo cannot solve the
problem of the framework, the optimal solution will be approximated using SA. This step was
consequently removed from the framework. The updated framework is included in Appendix F.
Step 8 Approximate the optimal layout using Simulated Annealing
A suitable programming language for coding algorithms such as SA, is RStudio. This program has many
packages to aid users. There is, for example, a package for SA called GenSA (Generalised Simulated
Annealing). However, it was found that this package does not allow for binary constraints to be added
in the solving process. Another solver called Optim was found that can approximate an optimal
solution for a non-linear function using SA. The RStudio code is included in Appendix E. Two library
tools are used in the code, namely PhonTools and Functional. A brief explanation of the functions
written in RStudio is now provided.

Main function: contains all the input variables, including the available space, number of departments,
relationship diagram, distance matrix, and fixed cost matrix. Additionally, the dimensions and possible
orientations of each department are stored in a vector called size_array_3D. The curry function in
RStudio was used to change functions into new ones that depends on less variables;

VectorToMatrix and MatrixToVector functions: are able to change vectors into matrices and matrices
into vectors without losing any data. The reason for this is that the QSCP contains matrices as input
variables;

BlockMatrixFull function: tests different rotations of each department and returns false if no more
rotations of departments are able to fit into the available space;

BlockMatrixImproved function: ensures that the departments do not overlap and that they are all
placed inside the available space;

PlaceDepartment function: places a department with a specific width and height in the available space
(called MatrixBlock) and returns false if it fails;

FunctionQSCP function: the objective function of the QSCP is coded in this function;

ChangeDepartments function: swaps two positions of departments around; and

GridPosition function: makes finding the positions of departments easier for the user of the program.
The RStudio program will return the best set of parameters found and the objective function value that
corresponds to this value.
5.7.4 EXAMPLES
Four examples of layouts generated by the framework developed in this study will now be discussed.
In order to arrive at a useful solution, 10,000 iterations were run in RStudio for these examples.
Example 1 has 19 hospital departments and ran for approximately 14 minutes. The other examples
have only 16 departments and the computation time was reduced by nearly 10 seconds. The
131
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
dimensions for each department were chosen to be the same for all four examples. This is also a
variable that the user of the framework can decide upon in addition to the orientation of each
department.
The first example includes 19 hospital departments in the layout and the relationships between these
are shown in Table 42.
HCU
Lab
ICU
Mort
NICU
Rehab
4 5 3 3 3
5 5 10 9 8
7 7 8 10 7
5 3 10 2 2
3 5 3 3 3
0 0 4 4 4
0 0 2 3 3
1 0 6 6 6
0 0 4 4 4
9 10 4 4 4
6 4 -3 -2 -2
5 4 0 5 5
3 3 1 3 3
3 0 3 4 4
0 2 5 5 5
0 0 3 5 5
0 0 0 0 0
0 0 0 0 10
0 0 0 0 0
Radio
CCU
-3
3
4
3
3
0
0
0
0
4
-3
4
3
0
0
0
0
0
0
Psych
4
-4
3
-4
5
4
4
3
4
-3
3
1
0
0
0
0
0
0
0
Emer
0 -2 0 5 -3 -3 5
5 3 5 5 5 -3 -2
5 0 8 2 10 -2 4
5 3 5 4 3 -4 -2
3 0 3 10 3 0 5
0 0 0 0 0 3 5
0 0 0 0 -3 3 4
0 0 0 0 7 0 5
0 0 0 0 3 3 6
0 0 0 0 0 -3 4
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
Phar
Laun
10
10
-2
8
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Kitch
Out
3 3 3
0 10 10
0 0 4
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
S&DU
Paed
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
OS
Obst
Admin
Obst
OS
Paed
Out
Laun
Kitch
S&DU
Phar
Emer
Psych
CCU
Rehab
Mort
Lab
Radio
NICU
ICU
HCU
Admin
TABLE 42: EXAMPLE 1 RELATIONSHIPS
The output of Example 1 is shown in Figure 22. The blue lines indicate the highest positive
relationships (assigned 10) and the red lines indicate the strongest negative relationships
(assigned -4). It is clear that the departments with the highest relationships are placed in close
proximity. The departments with the most negative relationships are separated with the exception of
Rehab and Paed. There are two possible explanations for this placement. Firstly, there are not enough
available blocks to place these two departments separate (investigated further in the third example).
Secondly, the positive relationships between 4 and 10 are rated higher and took precedence.
132
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
1
2
25
26
49
50
73
74
Admin
Radio
10
Emer
10
Out
Laun
10
10
Osbt
10
254
10
278
Phar
10
OS
CCU
-4
10
10
401
ICU
NICU
449
356
379
380
403
404
Psych
Lab
Kitch
Paed
-4
332
355
HCU
-4
10
308
331
425
10
Rehab
402
307
S&DU
Mort
FIGURE 22: EXAMPLE 1 OUTPUT
The second example includes 16 hospital departments in the layout and the relationships between
these are shown in Table 2. In this example only positive relationships were included in the input.
5
0
4
0
5
5
4
5
6
4
0
0
0
0
0
0
4
0
3
0
5
4
4
3
4
0
3
1
0
0
0
0
0
3
4
3
3
0
0
0
0
4
0
4
3
0
0
0
Radio
Mort
0
0
0
0
0
3
3
0
3
0
0
0
0
0
0
0
Lab
Rehab
0 5 0
5 5 5
8 2 10
5 4 3
3 10 3
0 0 0
0 0 0
0 0 7
0 0 3
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
CCU
Emer
0
3
0
3
0
0
0
0
0
0
0
0
0
0
0
0
Psych
0
5
5
5
3
0
0
0
0
0
0
0
0
0
0
0
Phar
10
10
0
8
0
0
0
0
0
0
0
0
0
0
0
0
S&DU
Laun
Kitch
3 3 3
0 10 10
0 0 4
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
OS
Out
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Paed
Admin
Obst
OS
Paed
Out
Laun
Kitch
S&DU
Phar
Emer
Psych
CCU
Rehab
Mort
Lab
Radio
Obst
Admin
TABLE 43: EXAMPLE 2 RELATIONSHIPS
4 5
5 5
7 7
5 3
3 5
0 0
0 0
1 0
0 0
9 10
6 4
5 4
3 3
3 0
0 2
0 0
The output for Example 2 is provided in Figure 23. It can be seen that the departments with the
highest relationships between them are placed in close proximity.
133
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
1
2
3
4
5
25
26
27
28
29
49
50
51
52
53
73
74
75
76
77
Admin
12
13
14
36
37
38
Emer
Mort
10
107
9
131
Out
Lab
10
10
10
Obst
254
10
10
278
Phar
307
331
OS
CCU
Psych
355
379
10
S&DU
432
Radio
456
Laun
469
Rehab
Paed
Kitch
480
498
499
504
522
523
528
546
547
552
570
571
572
573
574
575
576
FIGURE 23: EXAMPLE 2 OUTPUT
The third example has 16 departments and the relationships between these departments are similar
to the relationships in the first example, as shown in Table 44. Take note that the positive relationships
are also similar to the second example, but negative relationships are now included.
Rehab
Mort
Lab
4 5
5 5
7 7
5 3
3 5
0 0
0 0
1 0
0 0
9 10
6 4
5 4
3 3
3 0
0 2
0 0
Radio
CCU
-3
3
4
3
3
0
0
0
0
4
-3
4
3
0
0
0
Emer
4
-4
3
-4
5
4
4
3
4
-3
3
1
0
0
0
0
Psych
0 -2 0 5 -3 -3 5
5 3 5 5 5 -3 -2
5 0 8 2 10 -2 4
5 3 5 4 3 -4 -2
3 0 3 10 3 0 5
0 0 0 0 0 3 5
0 0 0 0 -3 3 4
0 0 0 0 7 0 5
0 0 0 0 3 3 6
0 0 0 0 0 -3 4
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
Phar
Laun
10
10
-2
8
0
0
0
0
0
0
0
0
0
0
0
0
S&DU
Out
3 3 3
0 10 10
0 0 4
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
Kitch
Paed
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
OS
Obst
Admin
Obst
OS
Paed
Out
Laun
Kitch
S&DU
Phar
Emer
Psych
CCU
Rehab
Mort
Lab
Radio
Admin
TABLE 44: EXAMPLE 3 RELATIONSHIPS
The output of the third example is shown in Figure 24. In the second example, Rehab and Paed were
placed adjacent but when assigned a negative relationship in example 3, they are placed apart. A
134
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
similar occurrence happened with Admin and Mort. It is thus clear that the departments with negative
relationships are placed further apart. Since Paed and Rehab are placed apart in the third example and
not in the first, it can be concluded that a larger site will lead to improved solutions.
1
2
3
25
26
27
49
50
51
58
73
74
75
82
10
Admin
10
11
12
13
14
34
Emer
Phar
10
9
Out
Lab
10
10
10
Obst
10
-3
CCU
OS
Radio 10
346
370
-4
418
399
400
401
423
424
311
312
334
335
336
357
358
359
360
425
Mort
Psych
443
-4
457
Paed
481
505
310
333
S&DU
394
442
309
Rehab
Laun
-4
Kitch
529
553
503
504
527
528
551
552
575
576
FIGURE 24: EXAMPLE 3 OUTPUT
The last example, investigates what happens when there is only one strong relationship (10 assigned)
and the other relationships are rated relatively lower. The relationship diagram is shown in Table 45
and the relationship between Radio and Rehab is rated the highest.
135
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
0
2
3
1
1
0
0
2
1
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
1
0
0
0
0
0
0
0
2
0
1
0
2
2
1
2
2
1
0
0
0
0
0
0
1
0
1
0
2
1
1
1
1
0
1
1
0
0
0
0
0
1
1
1
1
0
0
0
0
1
0
1
1
0
0
0
1 2
2 2
2 2
2 1
1 2
0 0
0 0
1 0
0 0
3 3
2 1
2 1
1 10
1 0
0 1
0 0
Radio
Lab
2
2
1
1
3
0
0
0
0
0
0
0
0
0
0
0
Mort
0
2
2
2
1
0
0
0
0
0
0
0
0
0
0
0
Rehab
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
CCU
0
2
2
2
1
0
0
0
0
0
0
0
0
0
0
0
Emer
3
3
0
2
0
0
0
0
0
0
0
0
0
0
0
0
Psych
Out
1
3
1
0
0
0
0
0
0
0
0
0
0
0
0
0
Phar
Paed
1
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
S&DU
OS
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Laun
Obst
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Admin
Obst
OS
Paed
Out
Laun
Kitch
S&DU
Phar
Emer
Psych
CCU
Rehab
Mort
Lab
Radio
Kitch
Admin
TABLE 45: EXAMPLE 4 RELATIONSHIPS
The output of the fourth example is shown in Figure 25. It can be seen that Rehab and Radio are placed
adjacent. In examples one to three these two departments were placed far apart. Thus, the user of the
framework is able to place two specific departments adjacent when a significantly higher relationship
is assigned.
1
2
3
4
25
26
27
28
49
50
51
52
73
74
75
76
Admin
Laun
Emer
Out
Phar
Obst
250
278
312
Lab
336
OS
CCU
Psych
360
373
384
408
Mort
Radio
10
Paed
Rehab
S&DU
FIGURE 25: EXAMPLE 4 OUTPUT
136
427
428
429
430
431
432
451
452
453
454
455
456
475
476
477
478
479
480
502
503
504
526
527
528
550
551
552
574
575
576
Kitch
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
5.8 Conclusion
In this chapter, the quantitative and qualitative layout design methods are integrated via embedding. It
was consequently found that this study follows a concurrent transformative research approach with
embedded design features.
The QSCP was chosen and adapted to model the problem of designing a rural hospital layout. The
selection was based on its objectives, assumptions, inputs, and outputs as well as qualitative design
considerations. SA was chosen as an adequate solution method for the QSCP.
Lastly, the generic rural layout design framework was developed using Excel VBA, and RStudio. A
program in Excel VBA was developed to capture inputs, estimate values, and ultimately provide the
areas and building cost of each hospital department. This program guides the user to adhere to the
requirements of the law. A program developed in RStudio is able to approximate the optimal solution
for the Excel VBA outputs using SA. The user is able to change variables of the Excel VBA and RStudio
program according to preferences. The framework is adjustable to suit the needs of the user with
regards to:

Size of the site and ability to change shape in a rectangular fashion;

Accuracy, i.e. the number of blocks the site is divided into;

Total number of hospital beds;

Target occupancy rate;

Which optional departments are included in the layout;

Number of beds in each department;

Size of each department;

Orientation and dimensions of each department;

Circulation space in each department;

Relationship between departments, i.e. degree to which departments should be placed in close
proximity to each other and which ones should be separated;

Size of each room within departments;

Which optional rooms are included in each department; and

Construction cost per square meter.
The framework calculates and recommends the following:

Total number of beds, dependent on the community’s population size and the occupancy rate;

Number of beds per department, according to benchmarking of five South African hospitals;

Total building cost of each department;

Circulation space;

Size of each room, according to the South African law requirements; and

Relationship between departments, according to literature research on hospitals.
137
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 5 | PROPOSED LAYOUT DESIGN FRAMEWORK FOR RURAL HOSPITALS
Four examples of the output of this framework are also discussed. It is found that the output of each
example clearly models the specifications of the layout (i.e. input variables).
The next chapter aims to verify and validate the developed layout design framework.
138
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6
VALIDATION AND VERIFICATION
“Validation can be expressed by the query ‘Are you building the right thing?’ and verification by ‘Are you
building it right?’”
Boehm (1981)
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
139
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
6.1 Introduction
In order to prove the accuracy of the developed framework, as well as its adequacy, the findings and
the framework must be verified and validated. Expert analysis is used to make the necessary
adjustments to the framework before applying it to a case study on Semonkong Hospital.
6.2 Verification
After conducting research, it is important to make a judgement of the research findings, i.e.
evaluation (Claesson, 2006). However, evaluation procedures are not always applicable depending on
the nature of the reality studied and the theories used (Svensson, 2003). Two important dimensions of
evaluation are verification and validation. Verification is concerned with the truth or accuracy and the
predictive power of theories, models, and methods whereas validation deals with their relevance and
meaningfulness (Warell, 2001). Truth in this context refers to the assessment of the ability of the
theoretical and practical result to be used to explain a real world problem.
Verification of this study is done in three ways. Firstly, the layout design questions in Section 1.2.1 are
considered. Table 73 of Appendix C explains how and where each question is addressed in this study.
Secondly, the research objectives of this study are examined. Table 74 of Appendix C explains how and
where each objective is addressed in this study. Thirdly, the outputs of the developed framework are
checked in order to make sure that the program functions as it was intended to and that the results are
as expected. This is clarified in detail in Table 75 of Appendix C.
6.3 Various routes to validation
There exist three commonly utilised routes to validating claims made by researchers, including
interviews with experts, application to case studies, and implementation (Mouton, 2011). Each
exhibits its own set of strengths and weaknesses.
Interviews are basically meetings in which information is obtained from the interviewee so as to
disprove or confirm the claims made by a researcher. The four forms of interviews described by
Mouton (2011)
include
structured
self-administered
questionnaires,
structured
telephone
interviewing, semi-structured focus group interviewing, and free attitude interviewing methods.
The second route to validating the framework is through conducting case studies. By applying the
framework to chosen case studies, the framework can be shown to be useful in real world
problems/scenarios.
The third and final route to validating the framework is through implementation. Although it is the
truest test of the workability of a concept, it requires the hospital to be built and evaluated which will
140
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
take a few years, and many resources so as to be proven applicable. As already established in
Section 1.7, the time constraint on this study makes it infeasible to test the framework practically
through implementation.
For the purposes of this study, it is therefore sufficient to perform expert analysis together with a case
study. Each of these are now addressed in turn.
6.4 Subject-matter expert analysis
The aim of this section is to validate the research, determine its shortcomings, and to make necessary
improvements to the framework before applying it to a case study. Furthermore, potential for future
research is identified.
6.4.1 THE PROCESS FOR VALIDATION VIA INTERVIEWS
Expert analysis in the context of this study takes the form of an interview based assessment in which
each interviewee (expert) gives his perspective as to the rationality of the findings and their
applicability to providing an answer to the problem situation, as well as to gain key insights into what
their motivations and recommendations are for each point in consideration.
When validating a hospital design framework the following aspects need to covered:

Accuracy: the calculations must be correct;

Adherence: all the applicable laws must be adhered to;

Realism of costs: the costs must be realistic;

Inclusiveness: the correct departments must be included;

Correctness: the correct rooms must be considered;

Realism of areas: the calculated areas for rooms must be realistic; and

Realism of solution: the final layout must be realistic.
These aspects form the basis of the validation questionnaire.
6.4.2 INTERVIEWEE PROFILE
This study focuses on three main areas/domains, namely layout design, construction planning, and
healthcare. Experts were chosen according to their areas of expertise, in the hopes that each will
provide key insights into the validity of the different aspects of this study. It is of course possible for
experts to exhibit expertise in more than one field, which is beneficial. The roles of each are now laid
out.
Healthcare architect
An architect is a person who “designs buildings and in many cases also supervises their construction”
(Stevenson, 2015). This person is able to provide input on the following:
141
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION

Whether the proposed layout is realistic;

Whether the applicable laws and regulations are adhered to in the recommended layout; and

Whether the summary of the details of the laws and regulations that are applicable to each department
are sufficient to provide guidance to the designer.
Quantity surveyor
A quantity surveyor is a person who “calculates the amount of materials needed for building work, and
how much they will cost” (Stevenson, 2015). This person is able to determine the following:

Whether the cost estimate is realistic; and

Whether the calculations that underpin this estimate are correct.
Healthcare expert or manager
A healthcare expert or manager will be able to evaluate the framework from a medical viewpoint. A
healthcare manager is a person who “provides leadership and direction to organisations that deliver
personal
health
services, and to
divisions, departments, units, or services within those
organisations” (Buchbinder & Thompson, 2010). This person is able to determine the following:

Whether the layout is practical in the context of a hospital;

Whether the flow-relationship diagram is realistic;

Whether the necessary departments are included; and

Whether the necessary rooms are included.
In order to get unbiased results, more than one expert from each area of expertise was interviewed.
6.4.3 INTERVIEWEE SUMMARY
A total of six subject matter experts (SMEs) were interviewed for the purpose of validating the
framework via interviews. The majority of the SMEs are employed by the Western Cape provincial
government in either the Department of Health or the Infrastructure and Technical Management Chief
Directorate.
Rural
community
x
Healthcare
x
x
x
x Layout design
x
FIGURE 26: THE INTERVIEWEES' AREAS OF EXPERTISE
142
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
Figure 26 shows the three areas of focus of this study along with the expertise of each interviewee.
Interviewees with expertise in more than one area are positioned where these areas overlap. It is clear
that the areas of healthcare and layout design are thoroughly covered by the validators. The rural
community area will be further validated in the application of the framework to a case study of a rural
hospital to make up for the deficiency in rural community expert interviewees.
6.4.4 VALIDATION QUESTIONS
By interviewing a number of experts, a varied perspective can be achieved. The questions in Table 46
were posed to each interviewee.
TABLE 46: VALIDATION QUESTIONS
Area
1
Healthcare
Rural
community
2
3
4
5
6
7
Construction
8
9
10
Laws and
regulations
11
12
General
13
14
Validation questions
Do you agree with the choice of departments included in the framework for designing a
district hospital?
Do you agree with the rooms included in each department?
Do you think the relationship diagram is realistic?
And do you think this is a useful way to decide upon the placement of each hospital
department relative to another?
Do you think the generated layout is practical from a healthcare point of view?
Do you think the generated layout will be able to support the healthcare needs of the
community or changed accordingly?
Do you think the framework can be useful for estimating the initial costs for a hospital that
you are considering building?
Do you think it can be useful for estimating the initial size of the hospital and its
departments?
Do you think the cost estimation is accurate enough to use during the initial planning phase
of developing a new hospital?
Do you think the methods used and calculations are appropriate for designing a hospital
layout?
Do you think the framework shows an accurate summary of the applicable laws and
standards?
Do you think the summary of these guidelines can be deemed useful for a person who
needs to design a hospital?
Do you think the framework is useful for generating an initial concept layout that architects
and other members of the design team can use as input for the design process?
Would you be prepared to use this framework as an aid in the design phase of a rural
hospital?
Are there any changes or additions to the framework that you think would be useful?
6.4.5 SUMMARY OF FEEDBACK FROM EXPERTS
After interviewing the experts, a document was sent to each one of them in order to gather their
opinions, general comments, and recommendations for improvement. A copy of this document can be
found in Appendix D.
Summary of validation question answers
Table 47 provides a short account of the feedback gained from the interviews with the experts.
143
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
TABLE 47 INTERVIEWEE FEEDBACK SUMMARY
Question
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Answer summary
All the interviewees agreed with the choice of departments for designing a rural hospital layout
included in the framework.
With the exception one interviewee, all the interviewees agreed with the rooms included in each
department. One interviewee recommends that the new IUSS data should be used to determine
which rooms to include in the layout.
Four interviewees agreed that the relationship diagram is realistic and that it is a useful way to
decide upon the placement of each hospital department relative to another. One interviewee did
not answer this question and another stressed that it is only useful for the initial stages of design.
Four out of six interviewees think that it is practical from a healthcare point of view. One
interviewee mentioned that it would be helpful to consider the influence of IT in the design.
Another stressed the importance of experience and precedent studies to support the initial
inputs.
All the interviewees agreed that the generated layout will be able to support the healthcare needs
of the community or changed accordingly.
Five of the interviewees agreed that the framework is useful for the estimation of the initial costs
of the hospital. One interviewee recommends comparing the results with the IUSS’s OOM
estimator.
Four out of six interviewees agreed that it can be useful for estimating the initial size of the
hospital and its departments.
All the interviewees agreed that the cost estimation is accurate enough to use during the initial
planning phase of developing a new hospital.
All the interviewees agreed that that the methods used and calculations are appropriate for
designing a hospital layout.
Four out of six interviewees agreed that the framework shows an accurate summary of the
applicable laws and standards. One interviewee recommends that the new IUSS data would be
useful to include in the summary.
All the interviewees find these guidelines useful for designing a hospital.
Four of the interviewees think that the framework is useful for generating an initial concept
layout that architects and other members of the design team can use as input for the design
process. One interviewee comments that the concept layout should be informed by broader
issues such as size, location and form of the site, and the re-use of facilities.
All the interviewees are prepared to use this framework as an aid in the design phase of a rural
hospital.
This question pertains to recommendations for the framework. Three interviewees agree that no
changes are necessary. One interviewee recommends including the impact of information
communication technology and hospital business processes. Another interviewee finds the
framework useful, but recommends that it should be tested before using it. Another interviewee
comments that it will be useful if it is fast to generate a layout, but recommends comparing the
outputs to other similar trusted methods.
In conclusion, each of the validation questions was answered positively, with a few minor
considerations for improvement. These will be addressed shortly before applying the framework to a
case study.
Adjustments to framework
Based on the answers to the validation questions, the following areas of improvement were deemed
necessary by the validators:
144
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION

Usage of the Infrastructure Unit Support Systems (IUSS) guidelines: The IUSS guidelines were up
until recently not formally approved by the government of South Africa and for this reason it was not
originally included in the framework. Some of these guidelines are still in the process of being approved
(gazetted). The IUSS refers to voluntary standards or guidance documents that have been prepared as
guidelines by the National Department of Health for the benefit of all South Africans. They are for the use
of public as well as private healthcare providers. These guidelines were developed from a range of
sources including local and international literature, expert opinion, practice, and expert group
workshops. This was then compiled into a discussion document which was afterwards released for
public comment. This feedback was consolidated into a new document which was again released for
comment and review. Further feedback was incorporated into a proposal document and submitted to
the National Department of Health. Once the government signs the document off, the documents have
been approved or gazetted. Since the IUSS guidelines are new and based on years of research, it will be
valuable to incorporate these guidelines into the framework. Consequently, these guidelines were
analysed and the necessary sizes and rooms were updated in Appendix A as well as the Excel VBA
program. This should make the estimation of the initial size and cost of the hospital more realistic and
ultimately make the layout more practical;

Analysis of information communication technology (ICT): Information communication technology
may play a significant role in the design of rooms within a department. However, the role of ICT on a
department level is significantly less. Since the analysis of rooms within departments falls outside the
scope of this study (as discussed in Section 1.7), the influence of ICT on the framework is considered
insignificant;

IUSS’s OOM (Order of Magnitude) estimator: The purpose of this estimator is to enable the South
African Department of Health to determine an order of magnitude budget for the medium term
expenditure framework over the financial years during which the project will be planned and
constructed (Steyn, 2014). Consequently, the framework includes not only the construction cost, but
also professional fees, health technology cost, and commissioning cost. It was found that the OOM also
uses a construction cost of R14,250/m2 to estimate the construction cost of a district hospital. However,
the size of a new district hospital is estimated to be 90m2 per bed. This value will vary from hospital to
hospital since the sizes of rooms and departments included in the layout are different. Therefore, it can
be argued that the developed framework is a better estimation for the initial size of a hospital
layout; and

Testing the framework: The framework is applied in the next section to a case study.
The framework has therefore been adjusted in line with the recommendations made by SMEs during
the interview based validation. The developed framework is now applied to a case study which forms
the second part of the validation process.
6.5 Case study
In order to show that the developed framework is useful in real world problems, it is now applied to
the case study of the Semonkong Hospital Project (SHP). Firstly, a background on the SHP is provided.
145
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
Secondly, the framework is applied and the results are discussed. Lastly, feedback from the architects
involved with the SHP is given.
6.5.1 THE SEMONKONG HOSPITAL PROJECT
This section provides background on the SHP and focuses specifically on the rational for a new
hospital, governance of the new hospital, the role of the new hospital, applicable laws and standards,
and construction and site planning.
Rationale for a new hospital
Within the Semonkong region of Lesotho there is currently a need for an optimally designed hospital
plan adequate to serve the general community. If there is an emergency, residents from this
community will have to travel approximately 115 kilometres to the nearest hospital which is located in
Maseru.
The
ability
to
do
so
is
further
restricted
by
the
availability
of
ambulances (Semonkong Hospital Project, 2014). The SHP was initiated by a few medical students
from the University of the Free State. These students found the remains of an abandoned hospital in
the mountains of Semonkong during July of 2005. As a result they have since decided to build a new
hospital to serve the rural community of Semonkong and call it the Jehovah Shammah Hospital (which
can be translated to City of the Lord). The fact that there is no such hospital anywhere near the
proposed site which has the required capacity further motivates the need for this hospital. The
problem of designing the layout of this hospital fits the criteria of the developed framework of this
study, namely a new district hospital that aims to serve the surrounding rural community.
Governance of a new hospital
The SHP is governed by a number of stakeholders, including members of the International Mission
Hospital Trust, members of the Semonkong Hospital Trust, nursing professionals of the local
healthcare clinics, project architects, members from the local community, and the advisory board of
the Trusts. The architects involved with the SHP are selected as the ideal validators for the developed
framework of this study.
The new hospital will function as a private hospital funded by the Trust and is financially independent
from the Government of Lesotho. A Memorandum of Understanding between the Ministry of Health
(Lesotho) and the Trust was signed in May of 2014. This grants the Trust permission to develop the
hospital and establish the parameters of cooperation between the Ministry and the Trust.
The role of the Jehovah Shammah Hospital
The main purpose of the Jehovah Shammah Hospital is to provide effective, affordable health care
services to the Semonkong community. These services are not intended to duplicate the services of the
current clinics, but rather to complement the health activities of the community by providing support
for patient referral as well as various administrative, educational, and technical activities. It is
146
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
therefore essential that training facilities and appropriate accommodation are provided for visiting
medical specialists, general practitioners, nurses, allied healthcare workers, and international
students.
Applicable laws and standards
The Jehovah Shammah Hospital must be constructed and operated in accordance with all the
applicable Lesotho laws and standards. However, Lesotho has no formal building regulations. An
architect of the Queen Mamohato Memorial Hospital in Maseru recommended the use of South African
building regulations along with consultations with the Lesotho authorities during the design phase.
The SHP identified the following documents and references for building regulations:

South African National Building Regulations SANS10400;

South African Occupational Health and Safely Act Regulations; and

South African Regulations Governing Private Hospitals and Unattached Operating Theatre Units (R158).
The South African Occupational Health and Safely Act Regulations are relevant to the detailed design
of a hospital but insignificant to the block layout design. The other two regulations are however
applicable to the block layout design and as such, both of these were included in the framework in
Section 4.2.
In conclusion, the relevant regulations are included in the framework as well as the new IUSS
guidelines which were deemed a valuable contributor in designing a hospital layout.
Construction considerations
The SHP identified the following design considerations for Jehovah Shammah Hospital:

High-performance solutions, but low-technical design solutions where possible;

Ease of use should be target of design;

Placement of elements and departments should result in an optimal interrelationship among
departments;

Creation of unity, an interflow and interaction of people working together; and

Economical in space use.
Additional to a new hospital, a veterinary facility, and staff accommodation are also included in the
construction plan of the SHP. The available site for these three facilities is shown in Figure 27.
147
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
FIGURE 27: THE SEMONKONG HOSPITAL PROJECT'S SITE SPACE
6.5.2 APPLICATION OF THE FRAMEWORK TO THE CASE STUDY
In this section the seven steps of the developed framework are applied to the Jehovah Shammah
Hospital.
Step 1 Inputs
The first step defines the characteristics of the hospital layout. The following factors are addressed in
this step: rural population size, occupancy rate, circulation space, construction cost, and optional
departments.
The SHP estimated the population size of Semonkong as 89,717 (Graham, 2016). This value was
entered into the program. The suggested values for the occupancy rate, circulation space, and
construction cost were selected (85%, 28%, and R14,250/m2 respectively). Lastly, no optional
departments were selected.
Step 2 Hospital beds
Since the developed framework’s calculations are based on South African population data, the relevant
values are changed to suit the population of Lesotho. Lesotho has a population of 1,741,000 (Lesotho
Bureau of Statistics, 2015). The following data relevant to Lesotho was found, namely the average
length of stay in Lesotho being 4.99 days, and the admissions per year being 47,865 (Graham, 2016).
Consequently, the admission rate per year is 1 in every 36.35 of the population (equal to 6.76
admissions per day). With these new values and the population size of Semonkong taken as being
89,717, the framework suggests 40 beds for the Jehovah Shammah Hospital. However, the SHP
predicts an increase in hospital admissions and population growth and plans to build a long term
hospital for 120 beds. This value was therefore selected for the next steps.
148
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
Step 3 Beds per department
This step recommends the number of hospital beds to the user and allows the user to select the
number of beds. The suggested number of beds, and selected number of beds for the Jehovah
Shammah Hospital are shown in Table 48.
TABLE 48: SUGGESTED AND SELECTED NUMBER OF BEDS FOR THE JEHOVAH SHAMMAH HOSPITAL
Department
Administration
Obstetrics
Operating Suite
Paediatric Unit
Outpatient Unit
Laundry
Kitchen
Sterilisation and Disinfection Unit
Pharmacy
Emergency and Casualty Unit
Surgical Unit
Chronic Care Unit
Rehabilitation Unit
Mortuary
Laboratory
Radiology
Total number of beds
Suggested number of beds
0
24
2
24
20
0
0
0
0
7
12
26
5
0
0
0
120
Selected number of beds
0
20
0
26
10
0
0
0
0
16
10
30
8
0
0
0
120
The SHP decided not to include an Acute Psychiatric Unit in their layout. They plan to use single rooms
in the hospital for psychiatric patients. Additionally, they plan to include a separate unit for a surgical
ward away from the other wards. This unit is further discussed in the next step.
Step 4 Relationship chart
This step determines which departments should be placed close together and which ones should be
separated. The SHP plans to include a surgical ward. This unit is different from the CCU in that it has a
very close relationship with the operating suite, ICU, and HCU (Van der Schyf & Fleming, 2014). A new
relationship diagram is suggested which was also selected for the Jehovah Shammah Hospital. This is
shown in Figure 28.
149
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
Administration
Obstetrics
Operating Suite
Paediatric Unit
Outpatient Unit
Laundry
Kitchen
Sterilisation and DisinfectionUnit
Pharmacy
Emergency and Casualty Unit
Surgical Unit
Chronic Care Unit
Rehabilitation Unit
Mortuary
Laboratory
Radiology
3
3
10 3
10 10
4 10 0
-2 5 -2
8 5 3 0
5 0 5 5
3 3 8 5 -3
0 5 2 5 5
0 3 4 10 -2 5
0 10 3 10 -2 4
0 0 3 -2 4 -4 -3
0 0 5 -2 3 3 4
0 -3 5 5 -4 4 5 5
7 4 5 5 3 7 5
3 5 4 4 3 5 7
6 5 4 0 3 3
4 6 3 0 0 5
4 4 0 0 0
0 -3 0 1 0
1 4 0 0
1 4 9 0
4 5 10
3 5 4
3 4
3 3
0
2
FIGURE 28: DEPARTMENT RELATIONSHIPS SUGGESTED AND SELECTED FOR THE JEHOVAH SHAMMAH HOSPITAL
Step 5 Hospital departments
The sizes of rooms in each hospital department were selected. Refer to Appendix A for the minimum
areas of each hospital room.
Step 6 Summary
This step shows a summary of the sizes (generated in step 5) and construction cost of each
department in the Jehovah Shammah Hospital. This is shown in Table 49.
TABLE 49: SUMMARY OF THE SIZES AND COSTS OF EACH DEPARTMENT IN THE JEHOVAH SHAMMAH HOSPITAL
Departments
Administration
Obstetrics
Operating Suite
Paediatric Unit
Outpatient Unit
Laundry
Kitchen
Sterilisation and Disinfection Unit
Pharmacy
Emergency and Casualty Unit
Surgical Unit
Chronic Care Unit
Rehabilitation Unit
Sum of room
sizes (in m2)
Circulation
space (in %)
Total area
(in m2)
202
306
65
174
451
102
153
89
834
792
228
910
210
20
30
19.5
30
20
10
25
25
15
14.6
30
30
15
242
437
333
249
564
113
204
118
981
927
325
1 300
247
150
Cost per
department
(in Rands)
3 448 500
6 227 250
4 745 250
3 548 250
8 037 000
1 610 250
2 907 000
1 681 500
13 979 250
13 209 750
4 631 250
18 525 000
3 519 750
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
SUMMARY OF THE SIZES AND COSTS OF EACH DEPARTMENT IN THE JEHOVAH SHAMMAH HOSPITAL (CONTINUED)
Mortuary
Laboratory
Radiology
Total area of hospital rooms
Total area of hospital
Total cost
53
27
180
4 774
-
15
15
15
-
62
31
209
6 342
-
883 500
441 750
2 978 250
90 373 500
Consequently, a total floor area of 6,342m2 is required and will cost approximately R90,373,500 to
build the Jehovah Shammah Hospital.
Step 7 Approximate the optimal layout using Simulated Annealing
Before using RStudio to approximate the optimal layout solution for the Jehovah Shammah Hospital, it
is necessary to consider the site space as well as the dimensions of each department. The areas of each
hospital department were chosen to be divided into 4 by 4 meter blocks. The resulting number of
blocks required is shown in Table 50. Suggestions for dimensions of each department are also shown.
TABLE 50: DIMENSIONS AND NUMBER OF BLOCKS REQUIRED FOR THE JEHOVAH SHAMMAH HOSPITAL
Departments
Administration
Obstetrics
Operating Suite
Paediatric Unit
Outpatient Unit
Laundry
Kitchen
Sterilisation and Disinfection Unit
Pharmacy
Emergency and Casualty Unit
Surgical Unit
Chronic Care Unit
Rehabilitation Unit
Mortuary
Laboratory
Radiology
Total area of hospital
Total number of blocks required
Area
(in m2)
Number of blocks
required
242
437
333
249
564
113
204
118
981
927
325
1 300
247
62
31
209
6 342
-
15.1
27.3
20.8
15.6
35.3
7.1
12.8
7.4
61.3
57.9
20.3
81.3
15.4
3.9
1.9
13.1
403
Dimensions
(in number of
blocks)
4x4
4x7
4x5
4x4
6x6
3x3
4x3
3x3
8x8
7x8
4x5
9x9
4x4
2x2
2x2
4x3
-
From the calculations in Table 50 it is clear that 403 blocks of 4 by 4 meters each are required for
building the hospital. The available site space will now be considered.
The site space for the SHP (previously shown in Figure 27) is obtained for three types of buildings,
namely the Jehovah Shammah Hospital, a veterinary facility, and staff accommodation. The 103 meter
reference line (from Figure 27) can be used to approximate the large rectangle (shown in Figure 29) as
151
20
1
59
40
21
2
98
79
60
41
22
3
99
80
61
42
23
4
100
81
62
43
24
5
101
82
63
44
25
6
102
83
64
45
26
7
103
84
65
46
27
8
104
85
66
47
28
9
105
86
67
48
29
10
106
87
68
49
30
11
107
88
69
50
31
12
108
89
70
51
32
13
109
90
71
52
33
14
110
91
72
53
34
15
111
92
73
54
35
16
112
93
74
55
36
17
113
94
75
56
37
18
114
95
57
38
19
80
39
78
134
115
96
173
154
135
116
97
212
193
174
155
136
117
213
194
175
156
137
118
214
195
176
157
138
119
215
196
177
158
139
120
216
197
178
159
140
121
217
198
179
160
141
122
218
199
180
161
142
123
219
200
181
162
143
124
220
201
182
163
144
125
221
202
183
164
145
126
222
203
184
165
146
127
223
204
185
166
147
128
224
205
186
167
148
129
225
206
187
168
149
130
226
207
188
169
150
131
227
208
189
170
151
132
228
209
190
152
133
76
77
58
153
211
192
210
230
231
232
233
234
235
236
237
238
239
240
241
242
243
301
282
263
244
302
283
264
245
303
284
265
246
285
266
247
171
191
172
229
281
262
304
300
323
261
361
342
280
341
322
299
360
279
321
260
359
340
298
320
297
358
339
259
319
278
357
338
277
318
258
356
337
296
317
257
355
336
276
316
295
335
275
354
256
315
294
353
334
255
314
274
352
333
293
313
254
351
332
273
312
292
331
291
350
253
311
272
330
271
349
252
310
290
329
251
348
270
309
289
347
328
250
327
308
269
346
288
307
287
345
326
249
325
306
268
286
344
267
305
248
324
456
437
418
343
417
380
455
436
399
416
379
454
435
398
415
378
453
434
397
414
377
452
433
396
413
376
451
432
395
412
375
431
394
450
374
411
393
430
373
449
392
410
391
429
372
448
371
409
390
447
428
370
427
408
389
446
369
407
388
445
426
387
425
406
368
444
367
405
386
424
385
443
366
404
365
423
384
442
364
403
383
441
422
363
421
402
382
440
381
401
362
420
439
419
152
400
Hospital.
438
120
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
326 by 120 meters. This rectangle can be divided into 81 by 29 blocks (2,430 blocks). However, only
403 blocks are necessary for building the hospital. Therefore, it was decided to use 500 blocks as the
available space for the hospital (indicated as the blue rectangle in Figure 29).
326
100
103
FIGURE 29: DIMENSIONS FOR THE SITE OF THE JEHOVAH SHAMMAH HOSPITAL
The 500 blocks for the hospital can be divided into 20 by 25 blocks. This is shown in Figure 30 along
with the possible locations (456 in total) of the centres of the departments of the Jehovah Shammah
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
80 m
100 m
1
20
39
58
77
96
115
13 4
15 3
17 2
19 1
2 10
229
248
267 286
2
21
40
59
78
97
116
13 5
15 4
17 3
19 2
2 11 2 3 0
249
268
287 306
3
22
41
60
79
98
117
13 6
15 5
17 4
19 3
2 12
2 3 1 2 50
269
288
307 326
4
23
42
61
80
99
118
13 7
15 6
17 5
19 4
2 13
232
2 51 2 70
289
308
327 346
5
24
43
62
81
10 0
119
13 8
15 7
17 6
19 5
2 14
233
2 52
2 71 2 9 0
309
328
347 366
6
25
44
63
82
10 1
12 0
13 9
15 8
17 7
19 6
2 15 2 3 4
2 53
2 72
291
3 10
329
348
367 386
7
26
45
64
83
10 2
12 1
14 0
15 9
17 8
19 7
2 16
235
2 54
2 73
292
3 11 3 3 0
349
368
387 406
8
27
46
65
84
10 3
12 2
14 1
16 0
17 9
19 8
2 17 2 3 6
2 55
2 74
293
3 12
3 3 1 3 50
369
388
407 426
9
28
47
66
85
10 4
12 3
14 2
16 1
18 0
19 9
2 18
237
2 56
2 75 2 9 4
3 13
332
3 51 3 70
389
408
427 446
10
29
48
67
86
10 5
12 4
14 3
16 2
18 1 2 0 0
2 19
238
2 57
2 76
3 14
333
3 52
3 71 3 9 0
409
428
447
11
30
49
68
87
10 6
12 5
14 4
16 3
18 2
201 220
239
2 58
2 77 2 9 6
3 15 3 3 4
3 53
3 72
391
4 10
429
448
12
31
50
69
88
10 7
12 6
14 5
16 4
18 3
202
221 240
2 59
2 78
297
3 16
335
3 54
3 73
392
4 11 4 3 0
449
13
32
51
70
89
10 8
12 7
14 6
16 5
18 4
203
222
241 260
2 79
298
3 17 3 3 6
3 55
3 74
393
4 12
4 3 1 4 50
14
33
52
71
90
10 9
12 8
14 7
16 6
18 5 2 0 4
223
242
261 280
299
3 18
337
3 56
3 75 3 9 4
4 13
432
4 51
15
34
53
72
91
110
12 9
14 8
16 7
18 6
205 224
243
262
281 300
3 19
338
3 57
3 76
4 14
433
4 52
16
35
54
73
92
111
13 0
14 9
16 8
18 7 2 0 6
225 244
263
282
301 320
339
3 58
3 77 3 9 6
4 15 4 3 4
4 53
17
36
55
74
93
112
13 1
15 0
16 9
18 8
207 226
245 264
283
302
321 340
3 59
3 78
397
4 16
435
4 54
18
37
56
75
94
113
13 2
15 1
17 0
18 9
208
227 246
265 284
303
322
341 360
3 79
398
4 17 4 3 6
4 55
19
38
57
76
95
114
13 3
15 2
17 1
19 0
209
228
285 304
323
342
361 380
399
4 18
4 56
247 266
295
305 324
343
362
381 400
325 344
363
345 364
4 19
438
382
401 420
439
383
402
421 440
365 384
403
422
441
385 404
423
442
405 424
443
395
425 444
437
445
FIGURE 30: POSSIBLE LOCATIONS OF CENTRES OF THE DEPARTMENTS OF THE JEHOVAH SHAMMAH HOSPITAL
Based on calculations of step 7, the RStudio program was used. The values of the input variables
include: 𝑛 equals 16, 𝑚 equals 456, 𝑓 refers to the relationship diagram in Figure 28, 𝑑 refers to the
distances between the locations (matrix with 207,936 elements), and 𝑡𝑖𝑗 refers to fixed cost of placing
departments (matrix with 7296 elements). The program was run for 10,000 iterations and took
approximately 10 minutes. The solution is shown in Figure 31. An objective function value of 12,234
was obtained.
153
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
1
20
2
21
3
22
4
23
39
58
77
S&DU
229
248
230
249
Laun
Mort
324
343
362
381
400
419
438
325
344
363
382
401
420
439
326
345
364
383
402
421
440
422
441
423
442
424
443
425
444
270
Emer
Rehab
10
10
OS
Kitch
Radio
-4
10
10
SWard
Paed
-4
293
Obst
10
CCU
10
Phar
Lab
Out
10
281
300
282
301
Admin
171
190
209
228
247
266
359
378
397
416
435
454
360
379
398
417
436
455
361
380
399
418
437
456
FIGURE 31: PROPOSED LAYOUT OF THE JEHOVAH SHAMMAH HOSPITAL
Table 51 show the relationships that have the highest positive ratings (a rating of 10 refers to
absolutely necessary). The distances of between the centroids of these departments are indicated as
well as their adjacency. Furthermore, the most important negative relationships are shown (a rating of
-3 indicates an undesirable relationship). It is clear that the most important relationships are adhered
to and the departments with negative relationships are placed apart.
TABLE 51: EVALUATION OF THE PROPOSED LAYOUT FOR THE JEHOVAH SHAMMAH HOSPITAL
Department
relationship
Admin-Out
Obst-Paed
Obst-Out
OS-Emer
OS-SWard
Out-Phar
Emer-Radio
Obst-OS
Emer-Lab
OS-S&DU
Paed-Out
S&DU-Emer
OS-Lab
OS-Radio
Phar-Sward
Phar-CCU
Priority
rating
10
10
10
10
10
10
10
10
9
8
8
7
7
7
6
6
Centroid
distance
(in
blocks)
8
5
6
8
4
8
7
14
16
5
9
7
20
11
8
21
Adjacent
Department
relationship
Priority
rating
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
No
No
Yes
No
Admin-SWard
Admin-Phar
Admin-Radio
Obst-Laun
Paed-Laun
Paed-S&DU
Paed-Lab
Out-SWard
Out-CCU
Out-Rehab
Out-Radio
Laun-SWard
Laun-CCU
S&DU-SWard
S&DU-CCU
SWard-Lab
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
154
Centroid
distance
(in
blocks)
22
14
15
12
17
12
11
14
13
16
11
22
15
9
26
16
Adjacent
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
EVALUATION OF THE PROPOSED LAYOUT FOR THE JEHOVAH SHAMMAH HOSPITAL (CONTINUED)
Admin-CCU
Obst-Emer
Obst-Lab
Obst-Radio
CCU-Lab
Obst-S&DU
Obst-Phar
5
5
5
5
5
5
5
11
10
6
5
7
17
14
Yes
Yes
Yes
Yes
Yes
No
No
OS-Laun
Admin-Emer
Admin-Mort
Kitch-Emer
Emer-Rehab
Obst-Rehab
Paed-Rehab
5
-3
-3
-3
-3
-4
-4
18
22
18
13
10
10
15
No
No
No
No
No
No
No
Another layout arrangement was generated (shown in Figure 32) using a larger site area in the hopes
of finding an improved solution without the constraint of the available site area. A site size of 806
blocks (750 possible locations for the centroids of the departments) was used. The program was run
for approximately 11 minutes and 10,000 iterations. An objective function value of 37,743 was
obtained which is an improvement upon the previous solution since the QSCP is formulated as a
minimisation problem. According to this solution, a site area of 84 by 100 meters is required for the
optimal layout of the Jehovah Shammah Hospital. The two layout solutions are very similar with the
exception of the administration department and the kitchen (outlined in red in Figure 32).
1
26
2
27
3
28
4
29
51
76
101
S&DU
301
326
302
327
Laun
Mort
426
451
476
501
526
551
576
601
626
651
676
701
726
427
452
477
502
527
552
577
602
627
652
677
702
727
428
453
478
503
528
553
578
603
628
653
678
703
728
479
504
529
554
579
604
629
654
679
704
729
480
505
530
555
580
605
630
655
680
705
730
481
506
531
556
581
606
631
656
681
706
731
482
507
532
557
582
607
632
657
682
707
732
633
658
683
708
733
634
659
684
709
734
635
660
685
710
735
636
661
686
711
736
637
662
687
712
737
638
663
688
713
738
639
664
689
714
739
640
665
690
715
740
354
Emer
Rehab
OS
Radio
383
SWard
Paed
Obst
CCU
Lab
Out
Phar
365
390
366
391
Kitch
Admin
641
666
691
716
741
467
492
517
542
567
592
617
642
667
692
717
742
468
493
518
543
568
593
618
643
668
693
718
743
319
344
469
494
519
544
569
594
619
644
669
694
719
744
320
345
370
395
420
445
470
495
520
545
570
595
620
645
670
695
720
745
21
46
71
96
121
146
171
196
321
346
371
396
421
446
471
496
521
546
571
596
621
646
671
696
721
746
22
47
72
97
122
147
172
197
322
347
372
397
422
447
472
497
522
547
572
597
622
647
672
697
722
747
23
48
73
98
123
148
173
198
223
248
273
298
323
348
373
398
423
448
473
498
523
548
573
598
623
648
673
698
723
748
24
49
74
99
124
149
174
199
224
249
274
299
324
349
374
399
424
449
474
499
524
549
574
599
624
649
674
699
724
749
25
50
75
100
125
150
175
200
225
250
275
300
325
350
375
400
425
450
475
500
525
550
575
600
625
650
675
700
725
750
FIGURE 32: RSTUDIO SOLUTION OF THE JEHOVAH SHAMMAH HOSPITAL USING A LARGER SITE AREA
155
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
6.5.3 FEEDBACK FROM THE SEMONKONG HOSPITAL PROJECT
Three architects involved with the SHP agreed to act as validators for this study. The validation
questions given in Section 6.4.4 were presented to them.
Summary of the validation question answers
Table 52 provides a short account of the feedback gained from the interviews with the experts.
TABLE 52 SEMONKONG HOSPITAL PROJECT INTERVIEWEES FEEDBACK SUMMARY
Question
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Answer summary
All the interviewees agreed with the choice of departments for designing a rural hospital layout
included in the framework.
All the interviewees agreed with the rooms included in each department.
All the interviewees agreed that the relationship diagram is realistic and that it is a useful way to
decide upon the placement of each hospital department relative to another.
One interviewee fully agrees that it is practical form a healthcare point of view. Another one
mentioned that it is a reasonable starting point for an architectural team to work with in
consultation with medical staff. Another interviewee commented that it is practical from a
relationship point of view but not necessary from a dimensional one.
Two interviewees agreed that the generated layout will be able to support the healthcare needs of
the community or changed accordingly. Another one did not provide an answer since the
interviewee believes that there are so many factors that support healthcare needs (not just the
layout) that it is nearly impossible to answer.
Two interviewees agreed that the framework is useful for the estimation of the initial costs of the
hospital. Another interviewee commented that if the framework generates a fair rate (with the
required inclusions/exclusions), it would be appropriate to use for an initial cost estimate.
All the interviewees agreed that it can be useful for estimating the initial size of the hospital and
its departments.
Two interviewees mentioned that it using the cost estimate of the framework during the initial
designing phase of a new hospital depends on how much tolerance you are willing to accept in
budgeting, but it could be useful in the very early design phases as long as it is understood to be a
rough indicator. Another interviewee commented that it could be useful if the newest 2016
figures are used in the calculations.
Two interviewees agreed that the methods used and calculations are appropriate for designing a
hospital layout. Another interviewee stressed that it can be useful to inform the designer, but
should not be used by itself to design the final layout.
All the interviewees agreed that the framework shows an accurate summary of the applicable
laws and standards.
All the interviewees find these guidelines useful for designing a hospital.
All the interviewees agreed that the framework is useful for generating an initial concept layout
that architects and other members of the design team can use as input for the design process.
All the interviewees are prepared to use this framework as an aid in the design phase of a rural
hospital.
Two interviewees agreed did not find any necessary changes or additions to make to the
framework. Another interviewee recommended more flexibility regarding the shape of the site.
In conclusion, the SHP architects responded in a positive manner to each of the validation questions.
156
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 6 | VALIDATION AND VERIFICATION
General comments and adjustments
According to one of the interviewees the latest figure on estimating costs is R25,737/m 2 (estimated in
March 2016). This new value is thus included in the framework. This interviewee further suggested
that a value of R28,835/m2 should have been used in the cost estimation since the site of the Jehovah
Shammah Hospital is remote and transportation of building materials will have a significant impact on
this value. The framework allows for such changes during Step 1. Thus, the new estimated cost of the
Jehovah Shammah Hospital is R182,871,570.
One of the architects stated that the framework deals with a conundrum every architect designing a
hospital faces and that some statistical backup to make decisions in terms of adjacencies are appealing.
All three architects of the SHP are prepared to use this framework as an aid in the designing phase of
the Jehovah Shammah Hospital and they find it useful for generating an initial concept layout.
6.6 Conclusion
In this chapter, the accuracy of the developed framework was verified, and validated through two
routes.
To begin with, it was determined that there exist three possible routes to validate the framework of
this study, namely expert analysis, application to a case study, and implementation. Implementation
was determined as infeasible and therefore the remaining two routes were followed.
Firstly, a subject-matter expert analysis was employed through conducting interviews with six experts
knowledgeable in the areas of layout design, rural communities, and healthcare. As a whole, all the
interviewees were supportive of the framework. According to their recommendations, the necessary
adjustments were made to the framework before validating it further.
The second route to validation involved applying the framework to the case study of the Semonkong
Hospital Project. Three architects involved with this project acted as further validators for the
framework. They responded in a positive manner and are prepared to use the results of the case study
as an input into the hospital that they are planning to build.
Thus both routes confirmed the validity of the framework and potential for further research were
identified, which is discussed in the final chapter.
157
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS
“Many discoveries are reserved for ages still to come, when memory of us will have been effaced.”
Seneca (2010)
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
158
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 7 | CONCLUSIONS AND RECOMMENDATIONS
7.1 Introduction
In this chapter, a summary of the research findings of each chapter are presented followed by a
discussion of the contributions of the study. Recommendations for future research based on the
foundations laid by this study are also discussed.
7.2 Research summary
The thesis opened in Chapter 1 with a discussion of the importance of designing a layout that is
functional and supports its operations. The purpose of this study was defined as the development of a
layout design decision-support framework and concept demonstrator for rural hospitals using mixed
methods. The research methodology involved a concurrent mixed methods approach. It was later on
determined that the quantitative and qualitative layout design methods can be integrated via
embedding. Thus the research follows a concurrent transformation approach with embedding design
features.
Chapter 2 provided an analysis of quantitative layout design methods used for modelling as well as
solving layouts. Seven popular layout models were identified, compared, and their mathematical
formulations presented. Exact methods and metaheuristics as a means to solve these models were
examined. It was found that metaheuristics are commonly used by researchers to approximate the
optimal solution of these layout models since the computation time increases significantly with the
problem size. Computer-based algorithms for constructing and improving layouts were also discussed.
The qualitative layout design methods and hospital considerations were presented in Chapter 3. Three
traditional qualitative approaches to solving the FLP were presented and analysed. It was found that
some of the steps of Muther’s SLP Procedure can be combined with a layout model analysed in
Chapter 2. An investigation was conducted into the most important design considerations applicable
to hospital layouts in literature. The results involved considerations related to: patient-centeredness,
efficiency, flexibility and expandability, sustainability, and therapeutic environments. The way in
which these considerations relate to the layout models presented in Chapter 2 was determined.
In order to understand the context of a rural hospital, the differences and similarities between urban
and rural hospitals were investigated in Chapter 4. The most important standards and regulations
associated with hospital design pertaining to the floor layout were found to be the National Core
Standards and Regulations 158. Using these regulations as well as other guides found in literature, the
minimum room areas and departments of a district hospital were identified. The specific difficulties
faced in rural settings were deliberated upon to arrive at what appears to be the most important
considerations for the layouts of rural hospitals: minimise costs and patient flow through the hospital.
159
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 7 | CONCLUSIONS AND RECOMMENDATIONS
The layout design decision-support framework and concept demonstrator for rural hospitals was
developed in Chapter 5. The most appropriate layout model and solution method were determined to
be the QSCP formulation and SA respectively. Using these methods and the minimum prescribed areas
for each hospital room and department, a user interface was developed using Excel VBA and RStudio.
These programs guide the user in adhering to the requirements of the applicable laws and guidelines
whilst approximating the optimal hospital layout according to the user’s preferences. Four examples of
the output of this decision-support framework were also discussed. It was found that the output of
each example clearly models the specifications of the layout (i.e. input variables).
In Chapter 6 the decision-support framework was verified and then validated via subject-matter
expert analysis and a case study on the Jehovah Shammah Hospital. A total of nine experts
knowledgeable in the areas of layout design, rural communities, and healthcare were interviewed. As a
whole, every validator responded positively to the framework and each one indicated that they are
prepared to use the framework as an aid in the initial design phase of a rural hospital. A few minor
recommendations were incorporated into the framework before applying it to the case study. The case
study received positive responses from architects who are prepared to use the results of the
framework.
7.3 Research contributions
The research contributes to both the academic literature and the design teams of rural hospitals. It
adds to the academic literature by applying quantitative and qualitative layout design methods to a
real world problem, i.e. designing the layout of a rural hospital. It was acknowledged in Chapter 2 that
very few quantitative layout design methods focused on hospital layouts and no study was found that
uses the QSCP formulation for modelling a hospital layout. It also adds to literature regarding a mixed
methods approach. The chosen quantitative and qualitative layout methods were integrated via
embedding. This resulted in the framework generating a layout based on ideal adjacencies or
preferences of the designer. These ideal adjacencies were suggested according to research conducted
on each hospital department. Thus, the relationship diagram generated in this study adds to hospital
layout design literature.
The design teams of rural hospitals can benefit from this research by using the design decisionsupport framework and concept demonstrator to generate a near optimal layout according to
preferences while taking the necessary laws into account. This study provides a summary of the South
African laws and standards applicable to layout design, the key hospital design considerations, and
design considerations specific to rural hospitals. Although the focus was on rural hospitals, the
framework is appropriate for designing a district hospital in an urban setting since all the necessary
laws and standards are taken into account and the user of the framework is able to change inputs
160
Stellenbosch University https://scholar.sun.ac.za
CHAPTER 7 | CONCLUSIONS AND RECOMMENDATIONS
according to their preferences. Lastly, the SHP benefits by the results of the application of the
framework to the Jehovah Shammah Hospital.
7.4 Recommendations for future research
In this study quantitative and qualitative layout design methods were used to generate near optimal
arrangements of departments of district hospitals. Using the foundations of this research, it is
suggested that future studies could involve the following:

Optimising the arrangement of rooms within each hospital department: The minimum areas and
rooms for each hospital department were already determined. This could be a valuable input for the
detailed design of a hospital. It is also suggested to adapt the framework to include corridors and fixed
points; and

Scaling the framework to other types of hospitals: Regional hospitals, provincial tertiary hospitals,
central hospitals, and specialised hospitals are also part of the healthcare delivery system. Further
research can involve adjusting the framework for multi-floor layouts.
161
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
162
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Aarts, E., Korst, J. & Michiels, W. 2014. Simulated annealing. Search methodologies, 12:265-285.
Abido, M. A. 2002. Optimal power flow using tabu search algorithm. Electric Power Components and
Systems, 30(5): 469-483.
Accident Compensation Corporation. 2011. The Accident Compensation Corporation. Patient handling
guidelines:
facility
design.
[Online].
Available:
https://www.google.co.za/.mnhospitals.org%2FPortals%2F0%2FDocuments%2Fptsafety%2Flift%2F
FGI_PHAMAwhitepaper_042710.pdf&usg=AFQjCNEFBjiTaeUtLBnnfxj4bj9VDcyBzA&sig2=KPmm3G9u
YdlPU_OaEvOlnQ&bvm=bv.139250283,d.d2s [2016, May 17].
Adams, E. 2010. A semidefinite programming model for the facility layout problem. Unpublished
doctoral dissertation. Waterloo: University of Waterloo.
Advameg
Incorporated.
2015.
Reference
for
business.
Layout
[Online].
Available:
http://www.referenceforbusiness.com/management/Int-Loc/Layout.html [2016, March 12].
Agarwal, N. & Srivastva, A. 2009. Operations research in service delivery. Operations Research Journal,
2(1):103-107.
Ahmad, A. R. 2005. An intelligent expert system for decision analysis and support in multi-attribute
layout optimisation. Unpublished doctoral dissertation. Waterloo: University of Waterloo.
Ahmad, A. R., Basir, O., Hassanein, K. & Azam, S. 2008. An Intelligent Expert Systems' Approach to
Layout Decision Analysis and Design under Uncertainty. Intelligent Decision Making: An AI-Based
Approach 40(1):321-364.
AIDS Foundation of South Africa. 2014. HIV/AIDS in South Africa. What is the prevalence of HIV in South
Africa? [Online]. Available: http://www.aids.org.za/hivaids-in-south-africa/ [2016, May 20].
Al-Hakim, L. A. 2001. A note on efficient facility layout planning in a maximally planar graph model.
Official Journal of the International Foundation for Production Research, 39 (7):1549–1555
Ali, S. L. 2015. Management and Development Center. Facility layout [Online]. Available:
http://www.mdcegypt.com/Pages/Operation%20Management/Production%20&%20Operation%20
Management/Facility%20Layout/Facility%20Layout%20Hybrid%20Layouts%20(Combination)/Faci
lity%20Layout%20Hybrid%20Layouts%20(Combination).asp [2015, March 30].
Armour, G. C. & Buffa, E. S. 1963. A heuristic algorithm and simulation approach to relative location of
facilities. Management Science Journal, 9(2), 294-309.
Arora, R. & Saxena. 2007. A linearization technique for solving the quadratic set covering problem.
Optimisation: A Journal of Mathematical Programming and Operations Research, 39(1):33-42.
163
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Arora, S., Malhotra, N. & Shanker, R. 2011. Enumeration technique for solving multi-objective
quadratic set-covering problem using goal programming. Opsearch Journal, 48(1):20-29.
Arora, S., Shanker, R. & Saxena, R. 2011. Relation between set partitioning and set covering problems
with quadratic fractional objective functions. Opsearch Journal, 48(3):247-256.
Asl, A. D. & Wong, K. Y. 2015. Solving unequal-area static and dynamic facility layout problems using
modified particle swarm optimization. Journal of Intelligent Manufacturing, 11(2):1-20.
Australian Institute of Health and Welfare. 2005. Rural, regional and remote health: information
framework
and
indicators.
Rural
health
series
[Online].
Available:
http://www.aihw.gov.au/publication-detail/?id=6442467780 [2016, July 4].
Bagust, A., Place, M. & Posnett, J.W. 1999. Dynamics of bed use in accommodating emergency
admissions: stochastic simulation model. British Medical Journal, 319: 155–158.
Bain, C. A., Taylor, P. G., McDonnell, G. & Georgiou, A. 2010. Myths of ideal hospital occupancy. Medical
Journal of Australia, 193(5): 311.
Baloyi, N. 2011. IUSS Online. Schedule B: Minimum physical and building service requirements for Health
Establishments
[Online].
Available: http://www.iussonline.co.za/2011/12/schedule-b-minimum-
physical-and-building-services-requirements-for-health-establishments/ [2015, March 16].
Barach, P. & Dickerman, K. N. 2007. World Health Design. We Shape Our Buildings [Online]. Available:
http://www.worldhealthdesign.com/We-Shape-Our-Buildngs.aspx [2015, November 20].
Basner, M., Babisch, W., Davis, A., Brink, M., Clark, C., Janssen, S. & Stansfeld, S. 2014. Auditory and nonauditory effects of noise on health. Lancet, 383(9925):1325–1332.
Bazaraa, M. & Goode, J. 1971. A cutting-plane algorithm for the quadratic set-covering problem.
Operations Research, 23(1):150-158.
Bazaraa, M. S. & Elshafei, A. N. 1979. An exact branch‐and‐bound procedure for the
quadratic‐assignment problem. Naval Research Logistics Quarterly, 26(1):109-121.
Bazaraa, M. S. & Sherali, H. D. 1980. Benders' partitioning scheme applied to a new formulation of the
quadratic assignment problem. Naval Research Logistics Quarterly, 27(1):29-41.
Bazaraa, M. S. 1975. Computerized layout design: a branch and bound approach. Adani Institute of
Infrastructure Engineering Transactions, 7(4):432-438.
Beach, E. 2011. SF-Gate. How much room does a shower stall require [Online]. Available:
http://homeguides.sfgate.com/much-room-shower-stall-require-103145.html [2016, January 10].
164
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Beckmann, M. & Koopmans, C. 1957. Assignment problems and the location of economic activities.
Econometrica, 25(1):53-76.
Benders, J. F. 1962. Partitioning procedures for solving mixed-variables programming problems.
Numerische mathematik, 4(1):238-252.
Bertsekas, D. 1999. Nonlinear Programming. Athena Scientific, 12(1):10.
Bertsekas, D. 2003.
Nonlinear Programming. Massachusetts Institute of Technology: MIT
OpenCourseWare Spring 2003 [Online]. Available: http://ocw.mit.edu [2015, July 1].
Bester, M. J. 2006. Design of an Automated Decision Support System for Scheduling tasks in a Generalised
Job-Shop. Master’s thesis. Stellenbosch: University of Stellenbosch.
Bhatwadekar, S. G., Kulkarni, M.H. & Thakur, H. M. 2015. A Literature Review of Facility Planning and
Plant Layouts. International Journal of Engineering Sciences & Research Technology, 4(3):35-42.
Bhuiyan, N. & Baghel, A. 2005. An overview of continuous improvement: from the past to the
present. Management Decision, 43(5):761-771.
Bianchi, L., Dorigo, M., Gambardella, L. M. & Gutjahr, W. J. 2009. A survey on metaheuristics for
stochastic combinatorial optimisation. Natural Computing: an international journal, 8(2):239-287.
Bidyarthy,
A.
S.
2012.
Solution
to
quadratic
assignment
problems
using
Ant
Colony
System. Optimisation Techniques Project, Mathematical department of Indian Institute of Technology,
17(1): 615.
Bland, J. & Dawson, G. 1994. Large-scale layout of facilities using a heuristic hybrid algorithm. Applied
Mathematical Modelling, 1(1): 500-503.
Blum, C. & Li, X. 2008. Swarm intelligence in optimisation. Swarm Intelligence 14(3):43-85.
Blum, C. & Mercle, D. 2008. Swarm intelligence: Introduction and applications. Natural Computing
Series, Springer.
Blum, C. & Roli, A. 2003. Metaheuristics in combinatorial optimisation: Overview and conceptual
comparison. ACM Computing Surveys, 35(3):268-308.
Blum, C. & Roli, A. 2008. Hybrid metaheuristics: an introduction. Hybrid Metaheuristics 13(5):1-30.
Blum, C., Puchinger, J., Raidl, G. R. & Roli, A. 2011. Hybrid metaheuristics in combinatorial optimisation:
A survey. Applied Soft Computing, 11(6):4135-4151.
Blum, C., Roli, A. & Sampels, M. 2008. Hybrid metaheuristics: an emerging approach to optimisation.
International journal of human-computer studies, 114(1):1-15.
165
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Boehm, B. W. 1981. Software engineering economics. Englewood Cliffs: Prentice-hall.
Bozer, Y. A. & Meller, R. D. 1997. A re-examination of the distance-based facility layout problem.
Independent Institute of Education transactions, 29(7):549-560.
Bozer, Y. A., Meller, R. D. & Erlebacher, S. J. 1994. An improvement-type layout algorithm for single and
multiple-floor facilities. Management Science, 40(7):918-932.
Brandon-Jones, A. 2015. Fixed Position Layout. Wiley Encyclopedia of Management. 10(1):31-43.
Branson, R.D. & Blakeman, T.C. 2013. Respiratory care. Inter- and intra-hospital transport of the
critically ill [Online]. Available: http://rc.rcjournal.com/content/58/6/1008.full [2016, March 1].
British Medical Association. 2011. The psychological and social needs of patients. London: BMA
marketing and publications.
Broekmann, R. & Steyn, D. 2014. Infrastructure Unit Support Systems administration and related
services. Health Infrastructure Norms Advisory Committee.
Buchbinder, S. B. & Thompson, J. M. 2010. Career opportunities in health care management:
Perspectives from the field. Burlington: Jones & Bartlett Publishers.
Buffa, E. S., Armour, G. C, & Vollman, T. E. 1964. Allocating Facilities with CRAFT. Harvard Business
Review, 42(2):136-158.
Buhrs, D. 2013. Infrastructure Unit Support Systems adult physical rehabilitation. Health Infrastructure
Norms Advisory Committee.
Bullock, R. 2015. A month with three initiates. Africa Geographic Magazine, 23(5):15.
Burkard, R. E., & Bönniger, T. 1983. A heuristic for quadratic boolean programs with applications to
quadratic assignment problems. European Journal of Operational Research, 13(4):374-386.
Burkard, R. E. & Stratmann, K. H. 1978. Numerical investigations on quadratic assignment problems.
Naval Research Logistics Quarterly, 25(1):129-148.
Burkard, R. E. 1984. Locations with spatial interaction: quadratic assignment problem. New York:
Academic Press.
Burkard, R. E., Cela, E., Pardalos, P. M. & Pitsoulis, L. S. 1999. The quadratic assignment problem. Graz:
Kluwer Academic Publishers
Burkard, R., Dell’Amico, M. & Martello, S. 2009. Assignment problems. Philadelphia: Library of congress
cataloguing-in-publication data.
166
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Business
Dictionary.
2016.
WebFinance.
Throughput
[Online].
Available:
http://www.businessdictionary.com/definition/throughput.html [2016, March 18].
Büyüksaatçi, S. & Baray, A. 2014. A Comparison of Three Nature-Inspired Metaheuristics for the
Facility Layout Problem. 1st International Conference on Engineering and Applied Sciences Optimisation,
4(1):40.
Caccetta, L. & Kusumah, Y. S. 2001. Tersedia. Graph theoretic based heuristics for the facility layout
design
problems
[Online].
Available:
http://www.esc.auckland.ac.
nz/Organisations/ORSNZ/conf34/PDFs/Kusumah.pdf [5 Mei 2006].
Calkins, M. & Cassella, C. 2007. Exploring the cost and value of private versus shared bedrooms in
nursing homes. The Gerontologist, 47(2):169-183.
Carr, R. F. 2011. Whole Building Design Guide a program of the National Institute of Building Sciences.
Hospital [Online]. Available: https://www.wbdg.org/design/hospital.php [2016, March 7].
Carr,
R.
F.
2011.
Whole
Building
Design
Guide.
Hospitals
[Online].
Available:
http://www.wbdg.org/design/hospital.php [2015, January 5].
Castillo, I. & Peters, B.A. 2003. An extended distance based facility layout problem. Official Journal of
the International Foundation for Production Research, 41(11):2451–2479
Cela, E. 1998. The quadratic assignment problem: Theory and algorithms. Graz: Kluwer Academic
Publishers.
Cela, E. 2013. The quadratic assignment problem: theory and algorithms. Springer Science & Business
Media.
Center for Health Design. 2004. Robert Wood Johnson Foundation. Evidence-based hospital design
improves
healthcare
outcomes
for
patients,
families
and
staff
[Online].
Available:
http://www.rwjf.org/en/library/articles-and-news/2004/06/evidence-based-hospital-designimproves-healthcare-outcomes-for-.html [2015, December 10].
Centre
for
Operational
Research
and
Logistics.
2015.
[Online].
Available:
http://www.port.ac.uk/centre-for-operational-research-and-logistics/research/exact-methodsheuristics-and-metaheuristics/ [2015, February 10].
Charnsethikul, P. 1988. An exact branch and bound algorithm for the general quadratic assignment
problem. Doctoral dissertation. Lubbock : Texas Tech University.
167
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Chaudhury, H., Mahmood, A. & Valente, M. 2004. The use of single patient rooms versus multiple
occupancy rooms in acute care environments. Coalition for Health Environments Research, 15(6):2034.
Chen, C. W. & Sha, D. Y. 2005. Heuristic approach for solving the multi-objective facility layout
problem. International Journal of Production Research, 43(21):4493-4507.
Cheng, M.Y. & Lien, L.C. 2012. Hybrid artificial intelligence-based PBA for benchmark functions and
facility layout design optimisation. Journal of Computing in Civil Engineering, 2(3):612-624.
Claesson, A. 2006. A configurable component framework supporting platform-based product
development. Chalmers University of Technology.
Clark, V. L. P. & Ivankova, N. V. 2015. Mixed methods research: A guide to the field. Thousand Oaks:
SAGE publications.
Coetzer, M. & Fleming, E. 2014. Infrastructure Unit Support Systems health facility guides adult critical
care. Health Infrastructure Norms Advisory Committee.
Coetzer, M. 2013. Infrastructure Unit Support Systems diagnostic radiology. Health Infrastructure
Norms Advisory Committee.
Coetzer, P. & Pascarel, N. 2012. The Donor Committee for Enterprise Development. Primary Health Care
At
The
Base
Of
The
Pyramid:
RTT’s
Unjani
Clinic
Model
[Online].
Available:
http://www.enterprise-development.org/wp-content/uploads/unjani_clinic_model_20120921-1.pdf
[2016, May 27].
Commander, C. W. 2003. A survey of the quadratic assignment problem, with applications. Doctoral
dissertation. Florida: Tallahassee: University of Florida.
Conradie, D. C. U & Steyn, C. 2014. IUSS Online. Space guidelines: professional service provider
instructions
for
quantity
surveyors
and
architects
[Online].
Available:
http://www.iussonline.co.za/index.php/docman/document/procurement-and-operation/89-spaceguidelines-qs-architect-instruction-gazetted/file [2015, March 17].
Conway, D. G. & Venkataramanan, M. A. 1994. Genetic Search and the Dynamic Facility Layout
Problem. Computers & Operations Research, 21(8):955-960.
Creswell, J. W. & Clark, V. L. P. 2009. Designing and conducting mixed methods research. Thousand Oaks:
SAGE publications.
Creswell, J. W. Clark, V. L. P. & Garrett, A. L. 2008. Advances in mixed methods research. Thousand Oaks:
SAGE publications.
168
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Creswell, John W. 2015. A concise introduction to mixed methods research. Thousand Oaks: SAGE
publications.
Cruz, C .J. J., Perez, J., Alejandro, A. & Terrén, R. D. 2013. Randomized Algorithms for Rich Vehicle Routing
Problems: From a Specialized Approach to a Generic Methodology. Doctoral dissertation. Internet
Interdisciplinary Institute (IN3), Faculty of Computer Science, Multimedia and Telecommunications.
Davis, C. 2010. The Ergolab. Measure twice, build once; informing hospital builds with ergonomic knowhow [Online]. Available: https://ergonomicedge.wordpress.com/2010/08/21/measure-twice-buildonce-informing-hospital-builds-with-ergonomic-know-how/ [2015, December 1].
De Jager, P. 2011. IUSS Online. Building engineering services, cost task group [Online]. Available: to
http://www.iussonline.co.za/2011/12/building-engineering-services-cost-task-group/ [2015, March
10].
De Jager, P. 2014. Infrastructure Unit Support Systems neonatal services. Health Infrastructure Norms
Advisory Committee.
De Jager, P. 2015. Infrastructure Unit Support Systems health facility guides tuberculosis services. Health
Infrastructure Norms Advisory Committee.
Deb, K. 1991. Optimal design of a welded beam via genetic algorithms. American Institute of
Aeronautics and Astronautics Journal, 29(11):2013-2015.
Department of Health, 2001. Limpopo government. Manual for the Planning of a Food Service Unit and
Dining Hall for a Hospital or Health Institution. [Online]: Available: http://docplayer.net/16717542Policy-for-food-service-management-in-public-health-establishments.html [2016, March 10].
Department of Health. 2004. Republic of the Philippines Department of Health. Guidelines in the
Planning
and
Design
of
a
Hospital
and
Other
Health
Facilities
[Online].
Available:
http://www.doh.gov.ph/system/files/planning_and_design_0.pdf. [2015, April 9].
Department of Public Works. 2001. South African Government. Standard Electrical, Mechanical and
Architectural
Guideline
for
the
Design
of
Accessible
Buildings
[Online].
Available:
http://www.publicworks.gov.za/PDFs/consultants_docs/disabled.pdf [2015, May 29].
Dewey, J. 2013. Logic: the theory of inquiry. Curriculum inquiry, 22(2):119-139.
Di Caro, G. & Dorigo, M. 1998. Ant colonies for adaptive routing in packet-switched communications
networks. Parallel Problem Solving from Nature, 18(3): 673-682.
169
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Directorate General of Health Services. 2012. National Health Mission. Indian Public Health Standards:
Guidelines
for
District
Hospitals
[Online].
Available:
http://clinicalestablishments.nic.in/WriteReadData/108.pdf [2016, May 1].
Division of Building Technology, 1995. Laundry and linen service: planning/design and operation.
Pretoria: CSIR.
Donaghey, C. E. & Pire, V. F. 1990. Solving the facility layout problem with BLOCPLAN. Doctoral
dissertation. Houston: University of Houston.
Dorigo, M. & Stützle, T. 2003. The ant colony optimisation metaheuristic: algorithms, applications, and
advances. Handbook of metaheuristics, 20(3): 250-285).
Dorigo, M. 1992. Optimisation, learning and natural algorithms. Doctoral dissertation. Milan: University
of Milan.
Dorigo, M., Di Caro, G. & Gambardella, L. M. 1999. Ant algorithms for discrete optimisation. Artificial
life, 5(2):137-172.
Dorigo, M., Maniezzo, V. & Colorni, A. 1996. Ant system: optimisation by a colony of cooperating
agents. Systems, Man, and Cybernetics, Part B: Cybernetics, INSTITUTE OF ELECTRICAL AND
ELECTRONICS ENGINEERS Transactions, 26(1):29-41.
Dowbor, T. P. & Zerger, S. 2014. Mixed Methods Designs. CRICH Survey Research Unit Methodology Bits,
1(4):1-5.
Dowsland, K. A. & Thompson, J. M. 2012. Simulated annealing. Handbook of Natural
Computing 11(29):1623-1655.
Drira, A., Pierreval, H. & Hajri-Gabouj, S. 2007. Facility layout problems: A survey. Annual Reviews in
Control, 31(2):255-267.
Ducker, T., Laing, I., Leaf, A. & Newmarch, P. 2004. British Association of Perinatal Medicine. Designing a
Neonatal
Unit
[Online].
http://www.bapm.org/publications/documents/guidelines/DesigningNNU_May2004b.pdf
Available:
[2015,
April 18].
Dweiri, F. & Meier, F. A. 1996. Application of fuzzy decision-making in facilities layout
planning. International Journal of Production Research, 34(11):3207-3225.
Eberhart, R. C. & Kennedy, J. 1995. A new optimiser using particle swarm theory. In Proceedings of the
sixth international symposium on micro machine and human science, 1(3):39-43.
170
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Econex. 2010. Econex Health Reform Note 4: Integration of the public and private sectors under a
National
Health
Insurance
(NHI)
system
in
SA.
[Online].
Available:
http://indicators.hst.org.za/healthstats/181/data [2016, February 2].
Elliot‐Schmidt, R. & Strong, J. 1997. The concept of well‐being in a rural setting: understanding health
and illness. Australian Journal of Rural Health, 5(2):59-63.
Elshafei, A. N. 1977. Hospital layout as a quadratic assignment problem. Journal of the Operational
Research Society, 28(1):167-179.
Escoffier, B. & Hammer, P. 2007. Approximation of the Quadratic Set Covering problem. Discrete
Optimisation, 4(1):378-386.
Eygelaar, J. E. & Stellenberg, E. L. 2012. Barriers to quality patient care in rural district
hospitals. Curationis Journal, 35(1):1-8.
Farmer, J. D., Packard, N. H. & Perelson, A. S. 1986. The immune system, adaptation, and machine
learning. Physica D: Nonlinear Phenomena, 22(1):187-204.
Ferenc, K. 2015. Hospitals & health networks. The key to clinical space design [Online]. Available:
http://www.hhnmag.com/Magazine/2015/Apr/interview-debbie-gregory-health-care-design [2015,
Mei 1].
Feyzollahi, M., Shokouhi, A., Yazdi, M. M., & Tarokh, M. 2009. Designing a model for optimal hospital
unit layout. Pejouhandeh, 14(4):1-30.
Fifer, R. M. 1989. Cost benchmarking functions in the value chain. Strategy and leadership, 17(3):18-19.
Finch, L. E. & Christianson, J. B. 1981. Rural hospital costs: an analysis with policy implications. Public
Health Reports, 96(5):423.
Fischl, G. 2006. Psychosocially supportive design in the indoor environment. Masters dissertation. Lule̊ a:
Lule̊ a University of Technology.
Fleming, E. 2014. Infrastructure Unit Support Systems health facility guides emergency services. Health
Infrastructure Norms Advisory Committee.
Flemming, E. 2014. Infrastructure Unit Support Systems outpatient facilities. Health Infrastructure
Norms Advisory Committee.
Flemming, E. 2014. Infrastructure Unit Support Systems paediatric and neonatal. Health Infrastructure
Norms Advisory Committee.
171
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Foulds, L. R., & Giffin, J. 1985. A graph-theoretic heuristic for minimizing total transport cost in
facilities layout. International Journal of Production Research, 23(6):1247-1257.
Fourie, C., Sheared, D. & Steyn, D. 2014. Infrastructure Unit Support Systems laundry and linen. Health
Infrastructure Norms Advisory Committee.
Franke Kitchen Systems. 2013. Franke. Hospital, Mortuary & Industrial Products: Product Guide 2013
[Online].
Available:
http://www.franke.com/content/dam/watersystems/za/en/brochures/WS%20Hospital%20Product
%20Guide%202013.pdf [2015, March 2].
Frontline Systems. 2015. Frontline Solvers: the leader in analytics for spreadsheets and the web. Mixedinteger and constraint programming [Online]. Available: http://www.solver.com/mixed-integerconstraint-technology [2015, March 14].
G. Suresh & S. Sahu. 1993. Multiobjective Facility Layout Using Simulated Annealing. International
Journal of Production Economics, 32(1):239-254.
Geem, Z. W., Kim, J. H. & Loganathan, G. V. 2001. A new heuristic optimisation algorithm: harmony
search. Simulation, 76(2):60-68.
Gendreau, M. & Potvin, J. Y. 2010. Handbook of metaheuristics. New York: Springer.
Gendreau, M. 2003. An introduction to tabu search. Annalysis of Operations Research, 16(2):37-54).
Springer US.
Genova, K. & Guliashki, V. 2011. Linear integer programming methods and approaches: a
survey. Journal of Cybernetics and Information Technologies, 11(1):1-29.
Gero, J. S. & Kazakov, V. A. 2006. Evolving degin genes in space layout planning problems. Artificial
Intelligence in Engineering, 1(1):163-176.
Gessert, C., Waring, S., Bailey-Davis, L., Conway, P., Roberts, M. & Van Wormer, J. 2015. Rural definition
of health: a systematic literature review. BMC public health, 15(1):1.
Giffin, J. W. & Foulds, L. R. 1987. Facilities layout generalized model solved by n-boundary shortest
path heuristics. European Journal of Operational Research, 28(1):382-391.
Gilmore, P. C. 1962. Optimal and suboptimal algorithms for the quadratic assignment problem. Journal
of the society for industrial and applied mathematics, 10(2):305-313.
Glover, F. 1986. Future paths for integer programming and links to artificial intelligence. Computers &
operations research, 13(5):533-549.
172
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Goldberg, D. E. & Deb, K. 1991. A comparative analysis of selection schemes used in genetic algorithms.
Foundations of genetic algorithms, 1(3):69-93.
Goldberg, D. E. & Samtani, M. P. 1986. Engineering optimization via genetic algorithm. Electronic
computation, 2(11):471-482.
Goronescu, F. F., McClean, S. I. & Millard, P. H. 2002. A queueing model for bed-occupancy management
and planning of hospitals, Journal of the Operational Research Society, 53(1):19-24.
Graham, G. 2016. Personal interview. 3 September, Cape Town.
Green, L. V. 2002. How many hospital beds? The Journal of Health Care Organization, Provision, and
Financing, 39(4):400-412.
Groover, M. P. 2007. Work systems and the methods, measurement, and management of work. Upper
Saddle River: Pearson Prentice Hall.
Gupta, K., Gupta, S. K., Kant, S., Chandrashekhar, R. & Satpathy, S. 2007. Modern trends in planning and
designing of hospitals: Principles and practice. New Delhi: Jaypee Brothers Publishers.
Hahn, P. M. & Krarup, J. 2001. A hospital facility layout problem finally solved. Journal of Intelligent
Manufacturing, 1(1):489-496.
Hall-Andersen, L. B. & Broberg, O. 2014. Integrating ergonomics into engineering design: the role of
objects. Applied ergonomics, 45(3):647-654.
Hanafi, S. 2001. On the convergence of tabu search. Journal of heuristics, 7(1):47-58.
Hansen, P. 1979. Methods of nonlinear 0-1 programming. Annals of Discrete Mathematics, 5(1):53-70.
Harper, P. R. 2002. A framework for operational modelling of hospital resources. Healthcare
management science, 5(3):165-173.
Harris, B., Goudge, J., Ataguba, J. E., McIntyre, D., Nxumalo, N., Jikwana, S. & Chersich, M. 2011.
Inequities in access to health care in South Africa. Journal of public health policy, 30(1):102-123.
Hartl, R. F. & Preusser, M. 2009. Layout and Design. Production and logistics, 2(3):1.
Hassan, F. H. & Tucker, A. 2010. Using cellular automata pedestrian flow statistics with heuristic
search to automatically design spatial layout in 22nd International Conference on Tools with Artificial
Intelligence, 1(2):32-37.
Hassan, M. M. & Hogg, G. L. 1987. A review of graph theory application to the facilities layout
problem. Omega, 15(4):291-300.
173
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Hassan, M.M.D., Hogg, G.L. & Smith, D.R. 1986. SHAPE: A Construction Algorithm for Area Placement
Evaluation. International Journal of Production Research 24(1):1283-1295.
Helm, S. A. & Hadley, S. W. 2000. Tabu search based heuristics for multi floor facility layout.
International Journal of Production Research, 38(2):365–383.
Heragu, S. S. & Alfa, A. S. 1992. Experimental analysis of simulated annealing based algorithms for the
layout problem. European Journal of Operational Research, 57(2):190-223.
Heragu, S. S. & Kusiak, A. 1987. The facility layout problem. European Journal of Operational Research,
29(1):229-251
Heragu, S. S. & Kusiak, A. 1991. Efficient models for the facility layout problems. European Journal of
Operational Research, 53(1):1-13.
Herman Miller. 1999. Graphic Standards Programming and Schematic Design. Emergency Department
[Online]. Available: http://www.facilityresources.ca/healthcare/emergency_department.pdf [2016,
May 1].
Herman Miller. 2006. Herman Miller Healthcare. Sound Practices: Noise Control in the Healthcare
Environment
[Online].
Available:
http://www.hermanmiller.com/content/dam/hermanmiller/documents/research_summaries/wp_So
und_Practices.pdf [2015, May 20].
Holland, J. H. 1973. Genetic algorithms and the optimal allocation of trials. SIAM Journal on Computing,
2(2):88-105.
Huang, S. & Irani, S. 2013. Quality Management 345 Class notes. Ohio State University. [Online].
Available:
http://webcache.googleusercontent.com/search?q=cache:dmNOYxgPNVkJ:www.lean.org/FuseTalk/F
orum/Attachments/PaFA_ToolForDesigningALeanHospital [2015, May 3].
Hughes, R. G., Reiling, J. & Murphy, M.R. 2008. Patient Safety and Quality: An Evidence-Based Handbook
for Nurses. Rockville: Agency for Healthcare Research and Quality.
Huisman, E. R. C. M., Morales, E., Van Hoof, J. & Kort, H. S. M. 2012. Healing environment: A review of
the impact of physical environmental factors on users. Building and Environment, 58(4):70-80.
Humphreys J. 1998. Rural Health and the Health of Rural Communities. Doctoral dissertation. Bendigo:
La Trobe University.
Hunt, J. M. & Sine, D. M. 2013. Design Guide for the Built Environment of Behavioural Health Facilities.
New Jersey: NAPHS.
174
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Ibrahim, A. 2012. A framework for genetic algorithm application in hospital facility layout design. The
IUP Journal of Operations Management, 1(2):16-22.
Ilott, I., Gerrish, K., Laker, S. & Bray, K. 2013. Naming and framing the problem: Using theories, models
and conceptual frameworks. Yorkshire: National Institute for Health Research.
Infrastructure Unit Systems Support. 2011. IUSS Online. Paediatric and Neonatal Workshop [Online].
Available:
http://www.iussonline.co.za/iuss/wp-content/uploads/2011/09/Neonatal-paediatric-
workshop-paediatrics.pdf [2016, March 20].
Infrastructure Unit Systems Support. 2014. IUSS Online. IUSS Health Facility Guides: Hospital Mortuary
Services
[Online].
Available:
http://www.iussonline.co.za/iuss/wp-
content/uploads/2013/03/2014_03_19-IUSS-Hospital-Mortuary-Services-proposal.pdf [2016, March
25].
Institute of Medicine. 2013. The National Academies of Sciences, Engineering, and Medicine. Crossing the
Quality
Chasm:
The
IOM
Health
Care
Quality
Initiative
[Online].
Available:
http://iom.nationalacademies.org/Global/News%20Announcements/Crossing-the-Quality-ChasmThe-IOM-Health-Care-Quality-Initiative.aspx [2015. May 10].
Intensive Care Society. 1997. Standards for Intensive Care Units. Université Catholique de Louvain
[Online].
Available:
http://www.md.ucl.ac.be/didac/hosp/architec/UK_Intensive_care.pdf
[2016,
March 1].
Inter-Agency Standing Committee. 2010. World Health Organization. Global Health Cluster suggested
set
of
core
indicators
and
benchmarks
by
category
[Online].
Available:
http://www.who.int/hac/network/global_health_cluster/iasc_global_health_cluster_core_indicators_9
apr10.pdf [2016, February 3].
International Federation of Red Cross and Red Crescent Societies. 2015. International Federation of
Red Cross and Red Crescent Societies. Tuberculosis facts at a glance [Online]. Available:
https://www.ifrc.org/en/what-we-do/health/diseases/tuberculosis/tuberculosis-facts-at-a-glance/
[2016, May 23].
Islier, A. A. 1998. A genetic algorithm approach for multiple criteria facility layout design. International
Journal of Production Research, 36(6):1549–1569
Jabareen,
Y.
R.
2009.
Building
a
conceptual
framework:
philosophy,
definitions,
and
procedure. International Journal of Qualitative Methods, 8(4):49-62.
Jabulani Rural Health Foundation. 2014. Zithulele hospital. Hospital and clinic services [Online].
Available: http://www.zithulele.org/hospital-services.html [2016, May1].
175
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Jacobs, F. R., Chase, R. B. & Chase, R. 2010. Operations and supply chain management. New York:
McGraw-Hill.
Jamison, D.T., Breman, J.G., Measham, A.R., Alleyne, G., Claeson, M., Evans, D.B., Jha, P., Mills, A. &
Musgrove, P. 2006. Disease control priorities in developing countries. World Bank Publications.
Joint Commission Resources. 2003. Care Delivery and the Environment of Care: a Teamwork Approach.
Oakbrook Terrace: Department of Publications.
Joseph, A. & Ulrich, R. 2007. Sound Control for Improved Outcomes in Healthcare Settings. Centre for
health design, 4(1):1-15.
Kaku, B.K. & Thompson, G.L. 1986. An exact algorithm for the general quadratic assignment problem.
European Journal of Operational Research, 23(3):382-390.
Karaboga, D. 2005. An idea based on honey bee swarm for numerical optimisation. Masters dissertation.
Kayseri: Erciyes University.
Karlen, M. & Fleming, R. 2016. Space planning basics. Hoboken: John Wiley & Sons Inc.
Karvonen, S., Korvenranta, H., Paatela, M. & Seppala, T. 2007. Production flow analysis: a tool for
designing a lean hospital. World Hospitals and Health Services, 43(1):28-31.
Kellert, S. R., Heerwagen, J. & Mador, M. 2011. Biophilic design: the theory, science and practice of
bringing buildings to life. New York: John Wiley & Sons.
Kennedy, J., Kennedy, J. F., Eberhart, R. C. & Shi, Y. 2001. Swarm intelligence. Burlington: Morgan
Kaufmann.
Kenny, A., & Duckett, S. 2003. Educating for rural nursing practice. Journal of Advanced Nursing, 44(6),
613-622.
Kim, J. Y. & Kim, Y. D. 1995. Graph theoretic heuristics for unequal-sized facility layout
problems. Omega, 23(4):391-401.
Kirkpatrick, S., Gelatt, C. D. & Vecchi, M. P. 1983. Optimisation by simmulated annealing. Science
Journal, 220(4598):671-680.
Kitchen Appliances 123. 2015. Kitchen Appliances 123. Refrigeration [Online]. Available:
http://www.kitchenappliances123.co.uk/shop/pc/catalog/fpd34rs-1_2153_detail [2015, May 22].
Kobus, R. L. 2008. Building type basics for healthcare facilities. Hoboken: John Wiley & Sons.
Konak, A., Kulturel-Konak, S., Norman, B. A. & Smith, A. E. 2006. A new mixed integer programming
formulation for facility layout design using flexible bays. Operations Research Letters, 34(6):660-672.
176
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Konz, S. & Johnson, S. 2004. Work design: occupational ergonomics. Scottsdale: Holcomb Hathaway,
Publishers.
Kooopmans, T. C. & Beckman, M. 1957. Assignment Problem and the Location of Economic Acitivities,
Econometrics, 25(1):53-76.
Koza, J. R. 1992. Genetic programming: on the programming of computers by means of natural selection.
Cambridge: MIT press.
Kuijstermans, A. C. M. 2014. Hospital bed planning. Master’s dissertation. North Brabant: Eindhoven
Technical University.
Kumar, S. A. & Suresh, N. 2009. Operations management. New Delhi: New Age International.
Kunders, G. D. 2008. Hospitals: Facility Planning and Management. New Delhi: McGraw-Hill Offices.
KwaZulu-Natal Department of Health. 2010. KwaZulu-Natal Department of Health. Pharmaceutical
Policy and System Development [Online]. Available: http://www.kznhealth.gov.za/pharmacy/districtdec2010.pdf [2016, May 1].
Lacksonen, T. A. 1997. Pre-processing for static and dynamic facility layout problems. International
Journal of Production Research, 35(2):1095–1106
Laguna, M., Barnes, J. W. & Glover, F. W. 1991. Tabu search methods for a single machine scheduling
problem. Journal of Intelligent Manufacturing, 2(2):63-73.
Lawler, E. 1962. The quadratic assignment problem. Unpublished doctoral dissertation. Cambridge:
Harvard University.
Lawler, E. L. 1963. The quadratic assignment problem. Management science, 9(4)586-599.
Leaning Forward. 2005. Leaning Forward Process Improvement. The Five Lean Principles [Online].
Available: http://www.leaningforward.co.uk/principles.htm [2015, April 5].
Lee, H. Y., Yang, I. T. & Lin, Y. C. 2012. Laying out the occupant flows in public buildings for operating
efficiency. Building and Environment, 51(7):231-242.
Lee, R. & Moore, J. M. 1963. CORELAP. Computerized Relationship Layout Planning, Jounal of Industrial
Engineering, 18(4):195-200.
Leech, N. L. & Onwuegbuzie, A. J. 2009. A typology of mixed methods research designs. Quality and
quantity, 43(2):265-275.
Leung, J. 1992. A graph theoretic heuristic for designing loop layout manufacturing systems. European
Journal of Operational Research, 57(3):243–252
177
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Life Healthcare. 2014. Life Healthcare. Our quality: quality scorecard [Online]. Available:
http://www.lifehealthcare.co.za/Quality/Quality_Scorecard.aspx [2015, March 29].
Ligget, R. S. 2000. Automated facilities layout: past, present and future. Automation in Construction,
1(1):197-215.
Lim, M. H., Yuan, Y. & Omatu, S. 2000. Efficient genetic algorithms using simple genes exchange local
search
policy
for
the
quadratic
assignment
problem. Computational
Optimisation
and
Applications, 15(3):249-268.
Litvak, E. 2003. Optimising patient flow: moving patients smoothly through acute care
settings. Cambridge: Institute for Healthcare Improvement.
Loiola, E. M., De Abreu, N. M. M., Boaventura-Netto, P. O., Hahn, P. & Querido, T. 2007. A survey for the
quadratic assignment problem. European Journal of Operational Research, 176(2):657-690.
Lourenço, H. R., Martin, O. C. & Stützle, T. 2010. Iterated local search: framework and applications.
Handbook of Metaheuristics, (12):363-397.
Makinana, A. 2015. Initiation: Government wants to do away with traditional surgeons. News24, 9
September: 15.
Management Study Guide. 2015. Management Study Guide. Facility layout: objectives, design and factors
affecting the layout [Online]. Available: http://www.managementstudyguide.com/facility-layout.htm
[2015, December 2].
Mandal, J. K., Maukhopadhyay, S. & Pal, T. 2016. Handbook of Research on Natural Computing for
Optimisation Problems. Hershey: IGI Global.
Maniezzo, V., Dorigo, M. & Colorni, A. 1995. Theory and methodology algodesk: an experimental
comparison of eight evolutionary heuristics apphed to the quadratic assignment problem. European
Journal of Operational Research, 1(2):188-204.
Matsuzaki, K., Irohara, T. & Yoshimoto, K. 1999. Heuristic algorithm to solve the multi-floor layout
problem with the consideration of elevator utilization. Computers and Industrial Engineering,
36(5):487-502.
McBride, R. & Yormark, J. 1980. An implicit enumeration algorithm for quadratic integer programming.
Management Search, 26(3):282-296.
Media Club South Africa. 2014. Media Club South Africa. Healthcare in South Africa
[Online]. Available: http://www.mediaclubsouthafrica.com/africa/34-democracy/developmentbg/10
2-healthcare [2015, May17].
178
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Medical Dictionary for the Health Professions and Nursing . 2012. [Online]. Available:
http://medical-dictionary.thefreedictionary.com/outpatient [2016, March 17].
Medical Practice and Licensing Sector. 2013. Healthcare Facilities Regulation: General Hospital.
Northern Emirates: Licensing Department.
Mehrotra, N., Syal, M. & Hastak, M. 2005. Manufactured Housing Production Layout Design. Journal of
Architectural Engineering, 11(1):25–34
Meller, R. D. & Gau, K. Y. 1996. The facility layout problem: recent and emerging trends and
perspectives. Journal of manufacturing systems, 15(5):351-366.
Meller, R. D., Narayanan, V. & Vance, P. H. 1998. Optimal facility layout design. Operations Research
Letters, 23(3):117-127.
Méndez, C. P. J. 2011. Development of a hybrid metaheuristic for the efficient solution of strategic
supply chain management problems: application to the energy sector. Unpublished master’s thesis.
Barcelona: University of Catalonia.
Mitchell, J. E. 1998. Cutting plane algorithms for integer programming. New York: Rensselaer
Polytechnic Institute.
Montreuil, B. & Laforge, A. 1992. Dynamic layout design given a scenario tree of probable
futures. European Journal of Operational Research, 63(2):271-286.
Montreuil, B. 1991. A modelling framework for integrating layout design and flow network design.
Material Handling, 90(2):95-115).
Montreuil, B., Brotherton, E. & Marcotte, S. 2002. Zone-based facilities layout optimisation,
in Proceedings of the Industrial Engineering Research Conference. Hebron: 1-14.
Montreuil, B., Ratliff, H.D. & Goetschalckx, M. 1987. Matching based interactive facility layout.
Transactions, 19(3):271-279.
Motaghi, M., Hamzenejad, A. & Riyahi, L. 2011. Optimisation of hospital layout through the application
of heuristic tehcniques in Shafa Hospital. Internation Journal of Management and Business, 1(2):133138.
Motaghi, M., Hamzenejad, A., Riahi, L. & Soheili Kashani, M. 2011. Optimisation of hospital layout
through the application of heuristic technique (diamond algorithm) in Shafa Hospital. International
Journal of Management and Business Research, 1(3):133-138.
Mouton, J. 2011. How to succeed in your Master's and Doctoral Studies. Pretoria: Van Schaik Publishers.
179
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Nahmias, S. & Cheng, Y. 2009. Production and operations analysis. New York: McGraw-Hill.
Nakrani, S. & Tovey, C. 2004. On honey bees and dynamic server allocation in internet hosting
centers. Adaptive Behavior, 12(3-4):223-240.
Natarajan, S. 2010. Hospital Supportive Service. New Delhi: Excel Books.
National Centre for Education Statistics. 2006. National Centre for Education Statistics. Postsecondary
Facilities
Inventory
and
Classification
Manual.
Circulation
Area
[Online].
Available: https://nces.ed.gov/pubs2006/ficm/content.asp?ContentType=Section&chapter=3&section
=2&subsection=5 [2016, May 3].
National Department of Health. 2011. Human Resources for Health South Africa. HRH Strategy for the
Health Sector 2012 - 2016 [Online]. Available: http://www.gov.za/documents/human-resourceshealth-south-africa-hrh-strategy-health-sector-201213-201617 [2015, April 7].
National Department of Health. 2013. Department of Health: Province KwaZulu-Natal. Licensing of
Private Health Care Facilities [Online]. Available: www.kznhealth.gov.za/R158.pdf [2014, December 2].
National Department of Health. 2014 Health Information Systems Program. District Health
Information System Database [Online]. Available: http://indicators.hst.org.za/healthstats/181/data
[2016, February 4].
National Health Service Confederation. 2012. National Health Service Confederation. Defining mental
health
services:
promoting
effective
commissioning
and
supporting
[Online].Available: http://www.nhsconfed.org/~/media/Confederation/Files/Publications/Documen
ts/Defining_mental_health_services. pdf [2015, December 1].
National Health Service Scotland. 2002. Health Facilities Scotland. Scottish Health Planning Note 20.
Facilities
for
Mortuary
and
Post-Mortem
Room
Services
[Online].
Available:
http://www.hfs.scot.nhs.uk/publications-1/property/scottish-health-planning-notes/ [2016, May 11].
Neufert, E. & Neufert, P. 2012. Neufert Architects’ Data. Fourth Edition. Chichester: Blackwell
Publishing Ltd.
Nice, J. 2014. Infrastructure Unit Support Systems pharmacy. Health Infrastructure Norms Advisory
Committee.
O’Kelly, M. 1987. A quadratic integer program for the location of interacting hub facilities. European
Journal of Operations Research, 32(1):393-404.
Office of the Federal Register. 2002. The Code of Federal Regulations of the United States of America.
Washington: United States Government Printing Office.
180
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Ontario Ministry of Education. 2006. Ontario Ministry of Education. Planning and Design Guidelines for
Child Care Centres [Online]. Available: www.edu.gov.on.ca/eng/parents/planning_and_design.pdf
[2016, May 7].
Operations Research Society of South Africa. 2014. The Operations Research Society of South Africa.
Operations
Research
[Online].
Available:
http://www.orssa.org.za/wiki/pmwiki.php?n=Main.WhatIsOR [2014, November 10].
Organisation for Economic Co-operation and Development. 2013. Health at a glance. Hospital beds
[Online].
Available:
http://www.oecd-ilibrary.org/sites/health_glance-2013-
en/04/03/index.html?itemId=/content/chapter/health_glance-2013-34-en&mimeType=text/html
[2016, January 23].
Passino, K. M. 2002. Biomimicry of bacterial foraging for distributed optimisation and control. Institute
of Electrical and Electronics Engineers Control Systems, 22(3):52-67.
Pham, D. & Karaboga, D. 2012. Intelligent optimisation techniques: genetic algorithms, tabu search,
simulated annealing and neural networks. Springer Science & Business Media.
Philippi, B. 2012. Philippi Quality Construction. Factors to Consider in Hospital Design and Construction
[Online]. Available: http://philippiqualityconstruction.com/tag/hospital-design/ [2016, February 10].
Phillip, P. J., Mullner, R., & Andes, S. 1984. Toward a better understanding of hospital occupancy
rates. Health care financing review, 5(4):53.
Pierdait, A. 2016. A Source of Design Reference Standards. Malé: Construction Délegation.
Pinelog.
2015.
Pinelog.
Hydrotherapy
Pools
[Online].
Available:
http://www.pinelog.co.uk/hydrotherapy-pools.html [2016, May 11].
Railoun, A. & Van der Schyf, E. 2014. Infrastructure Unit Support Systems health facility guides mental
health. Health Infrastructure Norms Advisory Committee.
Rajeev, S. & Krishnamoorthy, C. S. 1992. Discrete optimisation of structures using genetic algorithms.
Journal of Structural Engineering, 118(5):1233-1250.
Ramjee, S. 2013. Comparing the Cost of Delivering Hospital Services across the Public and Private
Sectors in South Africa. Unpublished master’s thesis. Cape Town: University of Cape Town.
Ramkumar, A., Ponnambalam, S. & Jawahar, N. 2008. A popoulation-based bubrid ant system for
quadratic assignment formulations in facility layout design. The International Journal of Advanced
Manufacturing Technology, 1(2):548-558.
Rardin, R. L. 1998. Optimisation in operations research. New Jersey: Prentice Hall.
181
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Rechel, B., Wright, S., Barlow, J. & McKee, M. 2010. Hospital capacity planning: from measuring stocks
to modelling flows. Bulletin of the World Health Organization, 88(8):632-636.
Reenen, T. 2014. Infrastructure Unit Support Systems mortuary. Health Infrastructure Norms Advisory
Committee.
Richards, A. & How, J. 2005. Mixed-integer programming for control. American Control Conference.
Piscataway: Institute of Electrical and Electronics Engineers: 2676-2683.
Ritzman, L.P., Bradford, J. & Jacobs, R. 1979. A multiple objective approach to space planning for
academic facilities. Management Science, 25(9):895-906.
Rodriguez,
D.
2009.
Everyday
Health.
Tuberculosis
Infection
[Online].
Available:
http://www.everydayhealth.com/tuberculosis/how-does-it-spread.aspx [2016, May11].
Romeijn, H. E. & Zenios, S. A. 2008. Introduction to the special issue on operations research in health
care. Operations Research, 56(6):1333-1334.
Romero, L. A., Zamudio, V., Baltazar, R., Mezura, E., Sotelo, M. & Callaghan, V. 2012. A comparison
between metaheuristics as strategies for minimizing cyclic instability in ambient intelligence. Sensors,
12(8):10990-11012.
Russel, R. S. & Taylor, B. W. 1999. Operations Management. Chigago: Prentice Hall.
Saghafian, S., Hopp, W. J., Van Oyen, M. P., Desmond, J. S. & Kronick, S. L. 2012. Patient streaming as a
mechanism
for
improving
responsiveness
in
emergency
departments. Operations
Research, 60(5):1080-1097.
Schenker, A., Kandel, A., Bunke, H. & Last, M. 2005. Graph-theoretic techniques for web content mining.
Singapore: World Scientific.
Schiffauerova, A. 2014. Facilities Design and Material Handling Systems 1-14. Unpublished class notes
(Facilities Design and Material Handling Systems 1-14). Montreal: Concordia University.
Seneca, L. A. 2010. Natural questions. Chicago: University of Chicago Press.
Seppanen, J. & Moore, J. M. 1970. Facilities Planning with Graph Theory. Management Science,
17(2):242-253.
Shahin, A. & Poormostafa, M. 2011. Facility layout simulation and optimization: An integration of
advanced quality and decision making tools and techniques. Modern Applied Science, 5(4):95.
Sharma, S. & Sharma, P. 2007. Step by Step Hospital Designing and Planning. New Delhi: Jaypee
Brothers Medical Publishers.
182
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Shields, P. M. & Tajalli, H. 2006. Intermediate theory: The missing link in successful student
scholarship. Journal of Public Affairs Education, 2(1):313-334.
Shields, P. M. 1998. Pragmatism as a philosophy of science: A tool for public administration. Public
Administration, 4(1):195-225.
Shouman, M. A., Nawara, G. M., Reyad, A. H. & El-Darandaly, K. 2001. Facility layout problem and
intelligent techniques: a survey. 7th International conference on Production Engineering, Design and
Control. Alexandria: 1-31.
Shrewsbury and Telford Hospital National Health System Trust. 2014. National Health System Trust.
Putting
patients
first
[Online].
Available:
http://www.sath.nhs.uk/working-with-us/Staff_Newsletters/ppf_quarterly/1405_spring2014/vision_
operating_plan.aspx [2015, March 1].
Simpson, A. R., Dandy, G. C. & Murphy, L. J. 1994. Genetic algorithms compared to other techniques for
pipe optimisation. Journal of water resources planning and management, 120(4):423-443.
Singh, S & Sharma, R. 2006. A review of different approaches to the facility layout problem. Journal of
Advanced Manufacturing Technology, 30(1):425-433.
Skorin-Kapov, J. 1994. Extensions of the tabu search adaptation to the quadratic assignment problem.
Computers and Operations Research, 21(8):855-865.
Smith, J. M. & MacLeod, R. 1988. A relaxed assignment algorithm for the quadratic assignment
problem. International Journal of Rock Mechanics and Mining Sciences, 26(3):170-190.
Solimanpur, M. & Jafari, A. 2008. Optimal solution for the two-dimensional facility layout problem
using a branch-and-bound algorithm. Computers & Industrial Engineering, 55(3):606-619.
Soriano-Meier, H., Forrester, P. L., Markose, S. & Arturo Garza-Reyes, J. 2011. The role of the physical
layout in the implementation of lean management initiatives. International Journal of Lean Six
Sigma, 2(3):254-269.
South African Health News Service, 2012. South African Health News Service. National antenatal HIV
and
syphilis
prevalence
survey
released
[Online].
Available:
http://www.health-
e.org.za/2012/12/11/2011-national-antenatal-hiv-syphilis-prevalence-survey-released/
[2016,
March 19].
Statistics South Africa. 2013. Statistics South Africa. Data from General Household survey [Online].
Available: http://www.statssa.gov.za/publications/P0318/P03182013.pdf [2016, March 10].
183
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Stevenson,
A.
2015.
Oxford
Dictionary
of
English.
Architect
[Online].
Available:
https://en.oxforddictionaries.com/definition/architect [2016, August 1].
Stevenson, A. 2015. Oxford Dictionary of English. Quantity surveyor [Online]. Available:
https://en.oxforddictionaries.com/definition/quantity_surveyor [2016, August 1].
Steyn, D. & Boltman, A. 2014. Infrastructure Unit Support Systems catering services for hospitals. Health
Infrastructure Norms Advisory Committee.
Steyn, D. & Sheard, D. 2004. Infrastructure Unit Support Systems sterile supply. Health Infrastructure
Norms Advisory Committee.
Steyn, D. 2014. Infrastructure Unit Support Systems training and resource centre. Health Infrastructure
Norms Advisory Committee.
Steyn. 2014. Infrastructure Unit Support Systems support services. Health Infrastructure Norms
Advisory Committee.
Storn, R. & Price, K. V. 1996. Minimizing the Real Functions of the ICEC'96 Contest by Differential
Evolution, in International Conference on Evolutionary Computation. Porto: 842-844.
Strong, J. & Elliot-Schmidt, R. 1997. The concept of well-being in a rual setting: understanding health
and illness. Australian Journal of Rural Health. Edition 5(2):59-63.
Suman, B. & Kumar, P. 2006. A survey of simulated annealing as a tool for single and multiobjective
optimisation. Journal of the operational research society, 57(10):1143-1160.
Suo, X. 2012. Facility Layout. INTECH Open Access Publisher.
Svensson, D. 2003. Towards Product Structure Management in Heterogeneous Environments. Doctoral
dissertation. Göteborg: Chalmers University of Technology.
Talbi, E. G. 2009. Metaheuristics: from design to implementation. Hoboken: John Wiley & Sons.
Talbi, E. G. 2009. Metaheuristics: from design to implementation. John Wiley & Sons, 74(1):1-16.
Tam, K. Y. 1992. A simulated annealing algorithm for allocating space to manufacturing cells.
International Journal of Production Research, 30(1):63.
Tam, K.Y. 1992. Genetic algorithms, function optimisation, and facility layout design. European Journal
of Operational Research 63(5):322–346
Tarawneh, W. 2014. Hospital Departments and Relationship. Hospital Zones & Departments [Online].
Available: https://www.pulse/20140601220246-65586394-hospital-departments-and-relationshippart-i [2016, March 11].
184
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Thai, A. B. 2007. Evaluation of using swarm intelligence to produce facility layout solutions. Doctoral
dissertation. Louisville: University of Louisville.
Tobias, C. L. 1986. Human factors aspects of a graph theoretic model for hospital facility layout. Master’s
Thesis. Arizona: University of Arizona.
Tompkins, J. A., White, J. A., Bozer, Y. A. & Tanchoco, J. M. A. 2010. Facilities planning. NewYork: John
Wiley & Sons.
Tseng, L. Y. & Liang, S. C. 2006. A hybrid metaheuristic for the quadratic assignment
problem. Computational Optimisation and Applications, 34(1):85-113.
Tuzkaya, G., Gülsün, B., Tuzkaya, U. R., Onut, S. & Bildik, E. 2013. A comparative analysis of metaheuristic approaches for facility layout design problem: a case study for an elevator
manufacturer. Journal of Intelligent Manufacturing, 24(2):357-372.
Ulrich, R. S. 2001. Effects of healthcare environmental design on medical outcomes, in Design and
Health: Proceedings of the Second International Conference on Health and Design. Stockholm: Svensk
Byggtjanst: 49-59.
United States’ General Services Administration. 2012. Circulation: Defining and Planning. General
Services
Administration
[Online].
Available:
http://www.gsa.gov/portal/mediaId/190511/fileName/Circulation_-_Defining_and_Planning_ [2016,
March 12]
University Hospitals Institute. 2013. University Hospitals Institute. Key initiatives and goals [Online].
Available:
http://www.uhhospitals.org/about/university-hospitals-institute-for-health-care-quality-
and-innovation/what-is-quality/key-initiatives-and-goals [2015, February 10].
Van der Schyf, E. & Fleming, E. 2014. Infrastructure Unit Support Systems adult in-patient services.
Health Infrastructure Norms Advisory Committee.
Van der Schyf, E. & Flemming, E. 2015. Infrastructure Unit Support Systems maternity care facilities.
Health Infrastructure Norms Advisory Committee.
Van Reenen, T. 2014. Infrastructure Unit Support Systems health facility guides facilities for surgical
procedures: surgery. Health Infrastructure Norms Advisory Committee.
Van Reenen, T. 2014. Infrastructure Unit Support Systems laboratories. Health Infrastructure Norms
Advisory Committee.
Wainer, J. & Chesters, J. 2000. Rural mental health: Neither romanticism nor despair. Australian Journal
of Rural Health, 8(3):141-147.
185
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Wanga, W., Zmeureanua, R. & Rivard, H. 2005. Applying multi-objective genetic algorithms in green
building design optimisation. Building and Environment, 1(1):1512-1525.
Warell, A. 2001. Design syntactics: a functional approach to visual product form – theory, models and
methods. Doctoral dissertation. Göteborg: Chalmers University of Technology.
Welgama, P. S. & Gibson, P. R. 1995. Computer-aided facility layout: a status report. International
Journal of Advanced Manufacturing Technology, 10(1):66-77.
White. R. D. 2007. Recommended Standards for Newborn ICU Design. Unpublished paper delivered at
the Seventh Consensus Conference on Newborn ICU Design. 1 February, Clearwater Beach.
Winston, W. L. & Goldberg, J. B. 2004. Operations research: applications and algorithms. Boston:
Duxbury Press.
Wits
Centre
for
Rural
Health
Strategy.
2008.
University
of
the
Witwatersrand
Johannesburg. Wits Centre for Rural Health Strategy. https://www.google.co.za/url?sa=t&rct=j&q=&esr
c=s&source=web&cd=1&cad=rja&uact=8&ved=0ahUKEwjl-MP1mrfQAhUEYyYKHajICYQQFggZMAA&
url=https%3A%2F%2Fwww.health-e.org.za%2Fwp-content%2Fuploads%2F2014%2F02%2FUpdate
d-Rural-Fact-Sheet-27-Nov-2013.pdf&usg=AFQjCNFl69Dj3R4qcgIbXMZzwAVGOEzn3A&sig2=7aSR3h
yCiSYcMCVBHVR2XQ&bvm=bv.139250283,d.d24 [2016, May 3].
Wolf, G. 2011. Facility location: concepts, models, algorithms and case studies. International Journal of
Geographical Information Science, 25(2):331-333.
World Health Organization. 2000. District health facilities: guidelines for development and operations.
Geneva: World Health Organisation publications.
Yang, T. & Kuo, C. 2003. A hierarchical AHP/DEA methodology for the facilities layout design
problem. European Journal of Operational Research, 147(1):128-136.
Yang, X. S. & Deb, S. 2014. Cuckoo search: recent advances and applications. Neural Computing and
Applications, 24(1):169-174.
Yeh, I. C. 2006. Architectural layout optimisation using annealed neural network. Automation in
Contruction, 1(2):531-539.
York,
L.
2008.
AIA
Best
Practices.
Lactation
room
design
[Online].
Available:
http://www.aia.org/aiaucmp/groups/secure/documents/pdf/aiap037226.pdf [2016, May 11].
Youssef, H., Sait, S. M. & Ali, H. 2003. Fuzzy simulated evolution algorithm for VLSI
placement. International Journal on Applied Intelligence, Special issue on Applied Metaheuristics,
12(3):92.
186
Stellenbosch University https://scholar.sun.ac.za
REFERENCES
Zimring, C., Joseph, A. & Choudhary, R. 2004. The role of the physical environment in the hospital of the
21st century: once-in-a-lifetime opportunity. Concord: Centre for Health Design.
187
Stellenbosch University https://scholar.sun.ac.za
APPENDICES
CHAPTER 2 CHAPTER 3
QUANTITATIVE LAYOUT QUALITATIVE LAYOUT
DESIGN METHODS DESIGN METHODS
2
3
1
CHAPTER 1
INTRODUCTION
4
APPENDICES
CHAPTER 4
RURAL VERSUS URBAN
HOSPITALS
X
8
5
7
6
CHAPTER 7 CHAPTER 6
CONCLUSIONS AND VALIDATION AND
RECOMMENDATIONS VERIFICATION
188
CHAPTER 5
PROPOSED LAYOUT
DESIGN FRAMEWORK FOR
RURAL HOSPITALS
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A: HOSPITAL LAYOUT CONSTRAINTS
TABLE 53: ADMINISTRATION CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
Chief executive officer's office
36
Meeting room
30
HR manager
36
x
Clinical manager
25
x
Nursing services manager
Admin manager
Financial manager
Staff restroom
Toilets
Store (stationery)
Store (general)
25
25
25
30
6
6
6
x
x
x
Waiting area
25
Registry
Cashier
80
9
x
References
Broekmann and Steyn (2014)
Department of Health (2004);
Broekmann and Steyn (2014)
Department of Health (2004);
Broekmann and Steyn (2014)
Department of Health (2004);
Broekmann and Steyn (2014)
Sharma and Sharma 2007)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
KwaZulu-Natal Department of
Health (2010); Department of
Health (2004); Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
TABLE 54: OBSTETRICS CONSTRAINTS
Requirements
Minimum
area (in m2)
Maternity unit
Reception and admissions desk
Security area
Wheelchair and trolley bay
Main waiting area
6
6
12
52
Child play area
12
Public ablutions
Assessment room with toilet and
shower
8
Optional
x
25
References
Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Fleming (2014)
Van der Schyf and Flemming
(2015)
Fleming (2014)
Van der Schyf and Flemming
(2015)
Van der Schyf and Flemming
(2015)
Van der Schyf and Flemming
(2015)
Unit manager's office
20
Doctors' offices
20
Staff overnight room with suite toilet
and shower (May be shared with
postnatal ward if small unit)
9
Dedicated obstetric operating
theatre
40
x
Van der Schyf and Flemming
(2015); Medical Practice and
Licensing Sector (2013)
30
x
Steyn (2014)
Training facilities (May be shared
with postnatal ward if small unit)
Staff room
Staff changing room
Fleming (2014)
20
12
Fleming (2014)
Fleming (2014)
189
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Staff shower
Staff toilet
2
2
Delivery room
22
Toilets
6
Shower
2
Nurses' station
14
Clean utility
Cleaner's room
Dirty utility room
12
4
9
X-ray equipment bay
12
IT room
Store (medicine, clean linen,
consumable and sterile stock, and
equipment)
Antenatal ward
4
Fleming (2014)
Fleming (2014)
Van der Schyf and Flemming
(2015)
Van der Schyf and Flemming
(2015)
Fleming (2014)
Van der Schyf and Flemming
(2015)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Assume similar to trolley bay (Van
der Schyf & Flemming, 2015)
Fleming (2014)
14
Broekmann and Steyn (2014)
Bed area
8.3
National department of health
(2013)
Van der Schyf and Flemming
(2015)
Assume similar to rest room (Van
der Schyf & Flemming, 2015)
Broekmann and Steyn (2014)
National department of health
(2013)
Flemming (2014)
Flemming (2014)
Toilets
6
Patients' lounge
12
Storage
6
Nurses' station
6
Consulting rooms
Counselling room
High-dependency unit
12
10
Bed area
22
x
Toilets
6
x
Shower
Nurses' station
Storage
Postnatal ward
2
14
6
x
x
x
Postnatal bed area
8.43
Kangaroo mother care bed unit (7.5
per bed)
7.5
Toilets (1 per 6 mothers)
6
Bath (1 per 6 mothers)
2
Well-baby nursery (1.5 per basinet)
35
Day lounge
12
Storage
Cleaner's room
Clean utility room
6
4
12
Day room
12
Dirty room
9
x
190
Van der Schyf and Flemming
(2015)
Van der Schyf and Flemming
(2015)
Fleming (2014)
Railoun and Van der Schyf (2014)
Broekmann and Steyn (2014)
National department of health
(2013)
Van der Schyf and Flemming
(2015)
Van der Schyf and Flemming
(2015)
Assume similar to shower
(Broekmann & Steyn, 2014);
(Beach, 2011)
Van der Schyf and Flemming
(2015)
Assume similar to rest room (Van
der Schyf & Flemming, 2015)
Broekmann and Steyn (2014)
Fleming (2014)
Fleming (2014)
Assume similar to rest room (Van
der Schyf & Flemming, 2015)
Fleming (2014)
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Kitchen
8
Laundry area
6
Store (clean linen and general)
14
Railoun and Van der Schyf (2014)
Assume similar to general store
(Van der Schyf & Flemming, 2015)
Van der Schyf and Flemming
(2015)
TABLE 55: OPERATING SUITE CONSTRAINTS
Minimum
area (in m2)
Requirements
Operating room
Minor
General
Trauma
Cath lab
Digital
General
endoscopy
Urology
Orthopaedic
Neurosurgery
and spinal
20
40
49
49
49
References
National department of health
(2013); Van Reenen (2014)
55
55
60
60
Reception
9
Unit manager's office
20
Staff change rooms (4 per person/8
per theatre)
Storage (personal protective
equipment)
Patient-holding bay
Resuscitation trolley bay
Recovery area
Staff rest room
Anaesthetic induction room
(optional)
Optional
x
4
Van Reenen (2014)
Van Reenen (2014); Broekmann
and Steyn (2014)
National department of health
(2013); Van Reenen (2014)
2
Van Reenen (2014)
9
2
12
25
x
16
x
Van Reenen (2014)
Van Reenen (2014)
Van Reenen (2014)
Broekmann and Steyn (2014)
National department of health
(2013); Van Reenen (2014)
Van Reenen (2014); Broekmann
and Steyn (2014)
Van Reenen (2014)
Van Reenen (2014)
Van Reenen (2014)
Van Reenen (2014)
National department of health
(2013); Van Reenen (2014)
Van Reenen (2014)
Storage (mobile equipment)
5
Blood store
Storage (special instruments)
Storage (pharmaceutical)
Scrub-up/gowning room
Set up room (preparation room) (or
20 for 2)
Sluice
5
9
9
11
12
9
TABLE 56: PAEDIATRIC UNIT CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
References
Paediatric Inpatient Facility
Bed/cot area
14
Flemming (2014)
Basinet area (0 to 4 months old)
8.2
Flemming (2014)
High Care Unit (8% of beds)
14.14
x
Flemming (2014)
Isolation bed
14.1
x
Flemming (2014)
191
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Waiting area
40
x
Playroom
12
x
Toilets (1 per 8 patients)
2
Bath (1 per 8 patients)
2
Shower (1 per 12 patients)
2
Sister's office
12
x
Fleming (2014)
Staff room
20
x
Fleming (2014)
Staff ablutions
2
Fleming (2014)
Staff lockers
6
Broekmann and Steyn (2014)
Counselling room
9
Flemming (2014)
Sluice room
9
Van Reenen (2014)
Treatment room
12
x
Flemming (2014)
Clean utility room
12
x
Flemming (2014)
Dirty utility room
Ward kitchen (increased by 1.5 m2
per 10 beds)
Stores (linen, toys, equipment,
medicine)
Treatment room
5
Flemming (2014)
4
Flemming (2014)
15
Fleming (2014)
Assisted bathroom
12
x
Broekmann and Steyn (2014)
Van der Schyf and Flemming
(2015)
Flemming (2014)
Assume similar to shower
(Broekmann & Steyn, 2014);
(Beach, 2011)
Fleming (2014)
Flemming (2014)
4.05
Department of Public Works (2001)
Cleaner's room
4
Patient kit room
4
Education area (optional)
30
Unit manager's office
12
Fleming (2014)
Assume similar to change rooms
(Van Reenen, 2014)
Assume similar to training room
(Steyn, 2014)
Fleming (2014)
Doctor's office
21
Nurses' station
14
Clinical work space
12
Fleming (2014)
Nursery bed
9
Flemming (2014)
Work area
6
Flemming (2014)
Assume mounted
installation (Franke Kitchen
Systems Ltd, 2013)
Flemming (2014)
x
Fleming (2014)
x
Railoun and Van der Schyf (2014)
Neonatal unit
Baby bath and work surface
0.83
Isolation room
Neonatal bay (for observation and
stabilisation)
Medicine preparation area
12.8
Central nurses' station
x
6
Flemming (2014)
1
Assume at least 1
Railoun and Van der Schyf (2014)
14
x
10.8
De Jager (2014); Ducker, Laing, Leaf
and Newmarch (2004); Kitchen
Appliances 123 (2015)
Sister's office
12
Fleming (2014)
Counselling room
9
Fleming (2014)
Multipurpose storeroom
6
Broekmann and Steyn (2014)
De Jager (2014)
Milk kitchen (required if more than
20 neonatal and paediatric beds)
Small kitchen
10.8
x
192
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Toilet (public/disabled)
5
Interview room
10
Clean utility
12
Fleming (2014)
Assume similar to counselling room
(Flemming, 2014)
Fleming (2014)
Dirty utility
9
Fleming (2014)
Storage (equipment, linen, surgical)
10
Staff changing area
6
Staff resting area
20
Staff ablutions
2
Doctor's office
21
x
Fleming (2014)
Overnight facilities for doctors
9
x
Fleming (2014)
Sister's office
12
x
Fleming (2014)
Bed area
8.43
x
Ablutions
2
x
Day room/lounge
25
x
8.43
x
Lounge/dining room
10
x
Assume similar to postnatal bed
National department of health
(2013)
Fleming (2014)
Shower
2
x
Fleming (2014)
Toilet
2
x
Laundry
6
x
Fleming (2014)
Assume similar to general store
(Van der Schyf & Flemming, 2015)
x
x
Fleming (2014)
Broekmann and Steyn (2014)
x
Flemming (2014)
Fleming (2014)
Kangaroo Mother Care Unit
Assume similar to postnatal bed
National department of health
(2013)
Fleming (2014)
Assume similar to rest room
(Broekmann and Steyn (2014)
Lodger mothers' facilities
Bed area
TABLE 57: OUTPATIENT DEPARTMENT
Requirements
Minimum
area (in m2)
Outpatient Day Unit
Waiting area with play area
Reception
Patient toilets and showers
Nurses' station
10
9
6
14
Pre-operative nurses base
14
Scrub-up and gowning
Operating room
11
20
Set-up room
9
Mobile X-ray equipment bay
Sluice room
Equipment store
Dirty utility
4
9
5
9
Equipment service room
4
Medical gas cylinder store
6
Optional
x
193
References
Flemming (2014)
Van Reenen (2014)
Broekmann and Steyn (2014)
Railoun and Van der Schyf (2014)
Assume similar to nurses’ station
(Railoun & Van der Schyf, 2014)
Van Reenen (2014)
Van Reenen (2014)
Assume similar to pre-discharge
area (Flemming, 2014)
Flemming (2014)
Van Reenen (2014)
Van Reenen (2014)
Flemming (2014)
Assume similar to equipment bay
(Flemming, 2014)
Assume similar to general store
(Broekmann & Steyn, 2014)
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Pre-discharge recovery area
Disabled ablution
Clean utility
Cleaner's room
Sterile pack store
9
5
12
4
14
Clean linen store
8
Switch room
Female staff change rooms
Male staff change rooms
Staff toilets
Staff lockers
Staff room
4
12
12
2
12
20
x
x
Seminar room
30
x
Unit manager's office
Sister's office
Triage area
Public ablutions
Maternity outpatients
Reception and admissions desk
Main waiting area
Security area
Wheelchair and trolley bay
Child play area
Public ablutions
12
12
8.5
8
Teaching area
30
x
Breastfeeding area
12
x
Staff room
Staff toilet
Office
20
2
21
x
Storage
6
Records
Specimen collection and testing area
Sluice room
Dirty utility
16
12
9
9
Clean linen store
8
Surgical store
6
Equipment store
5
Medical store
6
Nurses' station
6
Mothers' lodging unit
Bed area
Bath (1 per 6 mothers)
Toilets (1 per 6 mothers)
Cleaner's room
Clean utility
9
40
6
12
12
8
x
x
7.5
x
2
x
2
4
12
x
x
x
194
Flemming (2014)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Steyn and Sheard (2004)
Van der Schyf and Flemming
(2015)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Assume similar to training room
(Steyn, 2014)
Van der Schyf and Fleming (2014)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Van Reenen (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Broekmann and Steyn (2014)
Coetzer and Fleming (2014)
Flemming (2014)
Assume similar to training room
(Steyn, 2014)
Assume similar to change room
(Flemming, 2014)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Assume similar to general store
(Van der Schyf & Flemming, 2015)
Broekmann and Steyn (2014)
Van Reenen (2014)
Van Reenen (2014)
Flemming (2014)
Van der Schyf and Flemming
(2015)
Assume similar to general store
(Van der Schyf & Flemming, 2015)
Van Reenen (2014)
Assume similar to general store
(Van der Schyf & Flemming, 2015)
National department of health
(2013)
Flemming (2014)
Assume similar to shower
(Broekmann & Steyn, 2014);
(Beach, 2011)
Flemming (2014)
Flemming (2014)
Flemming (2014)
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Day room
Dirty utility (sluice)
Kitchen and laundry area
25
x
9
10.8
x
x
Store (linen and general)
10
x
Assume similar to rest room
(Broekmann and Steyn (2014)
Flemming (2014)
De Jager (2014)
Van der Schyf and Flemming
(2015)
TABLE 58: LAUNDRY CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
Receiving/holding area
Sorting/pre-wash area
29
16
Washing-extractors
49
x
Tumble drying
Flat work ironers
Iron presses
Supervisors office
Sewing/repairs
Detergent store
37
57
33
10
20
7
x
x
x
Clean linen store
45
Staff facilities
25
x
x
References
Fourie, Sheared and Steyn (2014)
Fourie et al. (2014)
Office of the Federal Register
(2002); Fourie et al. (2014)
Fourie et al. (2014)
Fourie et al. (2014)
Fourie et al. (2014)
Fourie et al. (2014)
Fourie et al. (2014)
Fourie et al. (2014)
Office of the Federal Register
(2002); Fourie et al. (2014)
Fourie et al. (2014)
TABLE 59: KITCHEN CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
Delivery and reception area
9
Storage areas
37
x
Preparation areas
74
x
Cooking area
Wash-up areas
Cafeteria (optional)
28
48
15
x
References
National department of health
(2013); Gupta, Gupta, Kant,
Chandrashekhar & Satpathy
(2007); Steyn and Boltman (2014)
National department of health
(2013); Gupta et al. (2007); Steyn
and Boltman (2014)
National department of health
(2013), Gupta et al. (2007); Steyn
and Boltman (2014)
Steyn and Boltman (2014)
Steyn and Boltman (2014)
Steyn and Boltman (2014)
TABLE 60: STERILISATION AND DISINFECTION UNIT CONSTRAINTS
Requirements
Minimum
area (in m2)
Waste disposal
4
Sluice
4
Receiving and sorting
9
Optional
References
National department of health
(2013), (Gupta, Gupta, Kant,
Chandrashekhar & Satpathy, 2007)
National department of health
(2013), (Gupta et al., 2007)
Steyn and Sheard (2004)
195
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Trolley wash and holding
Store
3
8
Decontamination area
37
Textile area
Inspection and packing
Sterilisation area
Sterile store
Steriliser plant room
Office
6
34
10
14
6
4
Steyn and Sheard (2004)
Steyn and Sheard (2004)
Gupta et al. (2007); Steyn and
Sheard (2004)
Steyn and Sheard (2004)
Steyn and Sheard (2004)
Steyn and Sheard (2004)
Steyn and Sheard (2004)
Steyn and Sheard (2004)
Steyn and Sheard (2004)
x
TABLE 61: PHARMACY CONSTRAINTS
Requirements
Minimum
area (in m2)
Pharmacy dispensary
38
Store medicine
Staff room and lockers
Patient waiting area
Box room
Bulk store (may be located
elsewhere)
Cleaner's store
Dirty utility room
Good receiving area
Office
10
18
16
6
Optional
100
References
KwaZulu-Natal Department of
Health (2010); Nice (2014)
Nice (2014)
Nice (2014)
Fleming (2014)
Steyn (2014)
KwaZulu-Natal Department of
Health (2010)
Fleming (2014)
Fleming (2014)
Steyn (2014)
Steyn (2014)
x
4
9
6
12
TABLE 62: EMERGENCY AND CASUALTY UNIT CONSTRAINTS
Requirements
Minimum
area (in m2)
Bed area
Resuscitation area
Calming room (safe room)
Consulting room
14.145
20.5
7
12
Reception/Info/Help desk
12
Public ablutions
8
Disabled toilet
5
Baby change area
6
Waiting area
52
Triage workstation
X-Ray bay (10 per bed)
Casualty admissions desk and
cubicle
Children's play area
Sub-waiting area
Trauma beds
Central nurses station
Counselling room
10
4
Optional
x
References
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Herman Miller (1999); Fleming
(2014)
Fleming (2014)
Department of Public Works
(2001); Fleming (2014)
Fleming (2014)
KwaZulu-Natal Department of
Health (2010); Fleming (2014)
Fleming (2014)
Fleming (2014)
21
x
Fleming (2014)
12
16
10
15
9
x
x
x
x
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
196
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Clean utility
Cleaner's room
Dirty utility
Treatment room
Store (equipment, general, clean
linen, surgical)
Seclusion room
Plaster room
12
4
9
12
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
58
Fleming (2014)
15
16
Procedure room
40
Patient bay (holding and recovery)
Medical head
Doctors' office (plan for 3)
Meeting room
Unit manager's office
Sisters' office
Overnight stay with bathroom for
doctors
Staff shower
18
24
21
45
16
12
x
x
x
Fleming (2014)
Fleming (2014)
Herman Miller (1999); Fleming
(2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
9
x
Fleming (2014)
Staff toilet
2
Staff rest room
Staff change room
IT room
20
12
4
2
Fleming (2014)
Department of Health (2004);
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
TABLE 63: ACUTE PSYCHIATRIC FACILITY CONSTRAINTS
Requirements
Minimum
area (in m2)
Inpatient Units
Nurses' station
14
12
Treatment room
6
Medicine room
Clean utility
Clinical administration
Consulting room
Counselling room
Stores (linen and general store)
Dirty utility
12
12
12
9
17
9
12
Seclusion rooms
20
Patient lounge (for at least 15
people)
9
Dining room
9
Quiet room
8
Ablutions
4
Kit room
Ward kitchen
8
4
Cleaner's room
4
Patient laundry
8
Optional
x
x
x
197
References
Van der Schyf and Fleming (2014)
Railoun and Van der Schyf (2014)
Assume similar size as general
store (Broekmann & Steyn, (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Van der Schyf and Fleming (2014)
Fleming (2014)
Fleming (2014)
Assume similar to staff rest room
(Railoun & Van der Schyf, 2014)
R 158; Railoun and Van der Schyf
(2014)
Assume similar size as dining room,
National department of health
(2013)
Fleming (2014)
Assume similar to change rooms
(Van Reenen, 2014)
Van der Schyf and Fleming (2014)
Railoun and Van der Schyf (2014)
Assume similar to cleaner's room
(Railoun & Van der Schyf, 2014)
Railoun and Van der Schyf (2014)
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Dirty utility and waste management
Staff rest room
Staff ablutions
Lecture rooms
Doctors sleep-over
Occupational Therapy Unit
20
12
2
9
14
x
x
5.02
x
Group/Interview room
Activity/Craft room
Relaxation/Therapy/Lecture room
9
30
30
x
x
x
Storage
2
x
Office space
Railoun and Van der Schyf (2014)
Broekmann and Steyn (2014)
Railoun and Van der Schyf (2014)
Fleming (2014)
Van der Schyf and Fleming (2014)
Railoun and Van der Schyf (2014)
Assume similar space to other
hospital offices (Department of
Health, 2004)
R 158
R 158
R 158
Assume minimum area is 2,
National department of health
(2013)
TABLE 64: CHRONIC CARE UNIT CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
References
Bed area
10
Singe rooms
Bathroom
Assisted shower
Day room
Treatment room
Clean utility
Cleaner's room
Dirty utility
Inpatient unit kitchen (increased 1.5
for every 10 beds)
15
6
6.5
12
15
9
8
8
Kunders (2008); Van der Schyf and
Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
8
Van der Schyf and Fleming (2014)
Nurses' station
6
Patient kit room
Sluice
Store (clean linen)
Store (consumables)
Store (equipment)
Staffroom
Staff toilet
Clinical staff office
6
10
8
9
12
15
3
9
Unit manager's office
12
Patient waiting area
Public disabled toilet
10
5
R 158, Van der Schyf and Fleming
(2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014);
Broekmann and Steyn (2014)
Van der Schyf and Fleming (2014)
Van der Schyf and Fleming (2014)
TABLE 65: REHABILITATION UNIT CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
References
Inpatient unit
Bed area
Kunders (2008); Department of
Health (2004)
7.43
198
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Ablution and toilet facilities for
patients (1/8 beds)
1.67
Clean utility room
5
x
Treatment room
10
x
Separate storage space (user
decides)
1
x
Ward kitchen (includes milk kitchen)
Increased by 1.5 for every 10 beds
above 20 beds
Staff toilet
Single rooms
Disabled ablution facility (1/8 beds)
Nurse station
Shower or bath 1/12 patients
Dirty utility room, cleaner's room,
and soiled linen and waste room
Hydrotherapy
(for spinal/cranial rehabilitation)
Pool
Change rooms and lockers
Wheel chair toilet
Occupational Therapy
One-to-one work room
Clean work room
Dirty work room
Cognitive room (clean work room,
dirty work room, and cognitive room
may be combined into a single room
of 30)
Splint room
Storage space
ADL kitchen
Family/group conference room
Clinical psychologist
Group therapy room (may be shared
with social worker)
Physiotherapy
National department of health
(2013)
4
1.67
Department of Health (2004)
National department of health
(2013)
Department of Public Works (2001)
National department of health
(2013), assume similar size as duty
station
Beach (2011)
National department of health
(2013)
10
4.05
6
1.05
9
24
x
8
x
4
x
10
10
National department of health
(2013)
10
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
10
6
10
20
National department of health
(2013)
20
10
Storage space
9
Gym area
45
Pinelog (2015)
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
10
One-to-one work room
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
Assume minimum area is 1,
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
National department of health
(2013)
Other rooms
199
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Dining-room/Lounge (minimum of
20 m² for 10 patients, and thereafter
1,5 m² for each additional patient
Emergency room
20
National department of health
(2013)
16
National department of health
(2013)
TABLE 66: MORTUARY CONSTRAINTS
Requirements
Minimum
area (in m2)
Optional
References
Ablution area
Shower facilities
Changing room
Storage space
Offices (interview, body receiving,
pathologist)
Body/bier room
3
3
8
8
Reenen (2014)
Reenen (2014)
Reenen (2014)
Reenen (2014)
9
Reenen (2014)
9
Viewing room
9
Autopsy room
27
Reenen (2014)
National Health Service Scotland
(2002); Reenen (2014)
Reenen (2014)
TABLE 67: LABORATORY CONSTRAINTS
Requirements
Minimum
area (in m2)
Waiting areas
Ablution areas
Staff changing room
Storage (equipment and samples)
Specimen reception
16
8
12
5
12
Storage and treatment of waste
10
Testing area
25
Optional
References
Fleming (2014)
Van Reenen (2014)
Van Reenen (2014)
Van Reenen (2014)
Van Reenen (2014)
Van Reenen (2014); Gupta et al.
(2007)
Van Reenen (2014); Gupta et al.
(2007)
TABLE 68: RADIOLOGY CONSTRAINTS
Requirements
Reception
Minimum
area (in m2)
Optional
12
Radiographic room
34.1
Fluoroscopy room
Patient toilet
Waiting area
48.75
3
40
x
Mammography X-ray-imaging room
18
x
Change cubicles
Sub waiting area
Processing and viewing area
Counselling room
General ultrasound room
6
16
9
9
16
x
x
x
200
References
Fleming (2014)
Coetzer (2013); Department of
Health (2004)
Coetzer (2013)
Reenen (2014)
Broekmann and Steyn (2014)
Department of Health (2004);
Coetzer (2013)
Broekmann and Steyn (2014)
Fleming (2014)
Reenen (2014)
Fleming (2014)
Coetzer (2013)
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Computed Tomography (CT)
scanning
Magnetic resonance imaging room
Control room
Staff rest room
Staff toilet
Unit manager's office
32.5
x
Coetzer (2013)
41.25
11
12
2
16
x
Radiologist's office
12
x
Radiographer's office
12
Cleaner's station
Dirty utility
IT room
Stores (equipment, consumables,
medicine, linen, and general)
4
9
4
Coetzer (2013)
Coetzer (2013)
Broekmann and Steyn (2014)
Fleming (2014)
Coetzer (2013)
Assume similar to sister's office,
Fleming (2014)
Assume similar to sister's office,
Fleming (2014)
Fleming (2014)
Fleming (2014)
Fleming (2014)
Assume similar to general store
size, Broekmann and Steyn (2014)
x
24
TABLE 69: NEONATAL INTENSIVE CARE UNIT CONSTRAINTS
Requirements
Crib area
Visitor ablution and toilet facilities
Minimum
area (in m2)
12.8
1.67
Clean utility room
5
Separate storage space (optional)
1
Staff toilet
Isolation cubicle
Breast feeding area
Optional
x
1.67
6
4.55
Nurse station
6
Dirty utility room, cleaner's room,
and soiled linen and waste room
9
References
Flemming (2014)
Department of Health (2004)
National department of health
(2013)
Assume minimum area is 1,
National department of health
(2013)
Department of Health (2004)
National department of health
(2013)
York (2008)
Assume similar size as duty station
National department of health
(2013)
National department of health
(2013)
TABLE 70: INTENSIVE CARE UNIT CONSTRAINTS
Requirements
Minimum
area (in m2)
Bed area
20
Waiting area for visitors
10
Visitor toilet facilities
Ward kitchen
Staff toilet
Optional
1.67
x
4
1.67
Isolation bed
25
Nurse station (1 per 8 beds)
6
201
References
Assume similar to size of other
waiting areas (KwaZulu-Natal
Department of Health, 2010)
Department of Health (2004)
National department of health
(2013)
Department of Health (2004)
National department of health
(2013)
Assume similar size as duty station
National department of health
Stellenbosch University https://scholar.sun.ac.za
APPENDIX A | HOSPITAL LAYOUT CONSTRAINTS
Dirty utility room and cleaner's
room
7
Equipment storage space (additional
1m2 per ICU bed over 8 ICU beds)
16
(2013)
National department of health
(2013)
Assume similar to size of
emergency department storage
(KwaZulu-Natal Department of
Health, 2010)
TABLE 71: HIGH CARE UNIT CONSTRAINTS
Requirements
Bed area
Ablution and toilet facilities for
patients (1/8 beds)
Minimum
area (in m2)
Optional
15.75
Coetzer and Fleming (2014)
1.67
Department of Health (2004)
Clean utility room
5
x
Treatment room
10
x
Separate storage space
1
x
Ward kitchen (Increased by 1.5 for
every 10 beds above 20 beds)
Staff toilet
Single rooms
1.67
20
x
Disabled ablution facility
4.05
x
Nurse station
Shower/bath
Dirty utility room, cleaner's room,
and soiled linen and waste room
Waiting area
References
4
6
1.05
9
10
202
National department of health
(2013)
National department of health
(2013)
Assume minimum area is 1,
National department of health
(2013)
National department of health
(2013)
Department of Health (2004)
Coetzer and Fleming (2014)
National department of health
(2013)
Assume similar size as duty station,
National department of health
(2013)
Beach (2011)
National department of health
(2013)
Coetzer and Fleming (2014)
Stellenbosch University https://scholar.sun.ac.za
APPENDIX B: NATIONAL CORE STANDARDS
There is currently very limited knowledge on the procedures or methods required in order to design
the optimal floor layout of a hospital for use in remote rural areas of South Africa. South African
healthcare provides from the most basic primary care services, offered free by the government, to
highly specialised health services in both the private and public sector. However, the public sector is
under pressure with low resources and high demands trying to deliver services to about 80% of the
population while the state only contributes 40% of all expenditure on health (Media Club South
Africa, 2014). This healthcare system is not only inaccessible to many citizens, but public health
facilities have suffered poor management, underfunding and deteriorating infrastructure. Serving
rural communities in particular is very difficult since resources are even scarcer. Each year about 1200
medical students graduate and only 3% of them end up working in rural communities (Wits Centre for
Rural Health Strategy, 2008). Yet, almost half of the population resides in rural areas. Rural
communities have specific challenges and characteristics which contribute to their poor health
outcomes. These include working conditions, access to medical care, personal health and physical
environment.
According to the National Department of Health (2011:18b) the public health sector is underresourced and serves more patients than the private sector of South Africa. In 2011 the National
Department of Health developed the National Health Insurance (NHI) policy in order to improve
health. The NHI is aims to ensure that everyone has access to quality healthcare regardless of their
socio-economic status. This will lead to changes in management systems, service delivery and
administrative systems. In preparation for implementation of the NHI, service delivery in South
African healthcare facilities needs improvement. Thus a national quality assurance program called the
National Core Standards (NCS) was launched to drive facility improvements and serve as a benchmark
for quality. All South African facilities, both in the private and public sectors, shall be required to
comply with these standards in the near future (National Department of Health, 2011:3b). The aim of
the NCS is therefore to ensure that all health facilities provide quality care. In order to evaluate health
facilities, an external audit team measures compliance to a set of standards.
The NCS is currently one of the focus areas in preparation for achieving the NHI. The NHI is a phased
project that will be completed in a period of 14 years. The first five years constitute phase one and will
include the actual demonstration of the key administrative and technical aspects of the NHI so as to
achieve smooth integration of the systems as they mature and new information is obtained. One of the
first activities of this phase lies in conducting the Public Health Facility Audit. The audit identifies
existing health infrastructure which includes facilities, technology and management capacity. This will
allow for planning on how to improve the current health system and increase its capacity and
effectiveness of service delivery (Health Systems Trust, 2013). The second phase, period 2016 to 2020,
203
Stellenbosch University https://scholar.sun.ac.za
APPENDIX B | NATIONAL CORE STANDARDS
will focus on further real-life demonstration and the contracting of independent providers. In this
period the NHI will progressively take over administration, service delivery and technical health
functions. Provincial branches of the NHI Fund will be established and the health workforce will be
increased. In the third phase, period 2021 to 2025, the NHI will reach its maturing state. Private
hospitals will be selectively accredited and contracted and the population will be registered (National
Department of Health, 2011:16a).
The NCS are not new or additional standards, but a combination of existing guidelines and policies that
set out the basic compulsory requirements and expectations for ensuring quality healthcare facilities
in South Africa (National Department of Health, 2011b:82). The main objectives of the NCS include the
following:

To define the quality of care that every health facility in South Africa should comply with and use as a
guide to managers, staff and the public

To provide a benchmark against which healthcare facilities can be measured and shortcomings and
strengths identified

To establish a national structure to certify healthcare facilities as compliant to the NCS
In order for a healthcare facility to meet these standards, staff and managers need to know how far
their current performance is from being compliant. Continuous self-assessments are thus required to
ensure conformation to these standards. An independent body executes external healthcare
establishment audits in order to remove biases. The body will periodically issue audit reports and
assess the degree of compliance to the NCS. An independent regulator will issue certificates regarding
the extent of compliance to the standards according to the law. The appropriate process to enforce
compliance should be followed (National Department of Health, 2011b:51).
TABLE 72: FACILITIES AND INFRASTRUCTURE DOMAIN OF THE NATIONAL CORE STANDARDS
Sub-domain
7.1 Buildings
and grounds
Standard
7.1.1 The building meets all
applicable regulations
7.1.2 Infrastructure is
appropriately used according to
level of care
7.1.3 Waiting areas are
convenient and provide adequate
shelter and seating for patients
7.1.4 Buildings are safe and
adequately maintained
7.1.5 The health establishment is
Criteria
7.1.1.1 The health establishment has been licensed
annually against the R158 or R187 regulations * only
applicable to the private sector
7.1.1.2 The health establishment complies with
infrastructure standards * only applicable to the public
sector
7.1.2.1 Available facilities are regularly checked to
ensure they are fit for purpose
7.1.2.2 The health establishment layout is planned or
adapted to ensure it meets service and patient needs
7.1.3.1 Waiting areas are appropriately located and
adequate for the number of patients using them
7.1.4.1 The health establishment holds regular,
documented and comprehensive inspections of its
physical facilities
7.1.4.2 Maintenance is carried out promptly and
efficiently by qualified personnel
7.1.5.1 All areas are adequately furnished and provide
204
Stellenbosch University https://scholar.sun.ac.za
APPENDIX B | NATIONAL CORE STANDARDS
organised, furnished and
equipped to meet patient needs
and comfort
7.1.6 Grounds are maintained to
be safe and orderly
7.2
Machinery
and utilities
7.2.1 Electrical power, water,
sewerage and other internal bulk
supply systems meet the needs of
the establishment
an acceptable environment for patient care
7.1.6.1 A regular maintenance program ensures
grounds are safe and attractive
7.1.6.2 All pedestrian and vehicular access routes are
maintained to ensure the smooth running of the health
establishment
7.1.6.3 Emergency vehicle access roads are kept clear
7.2.11 Site and floor plans show the location and layout
of the main services (e.g. water, sanitation and
electricity)
7.2.1.2 Routine and emergency electrical power
services meet the needs of the health establishment
7.2.1.3 Routine and emergency water supplies meet the
needs of the health establishment
7.2.1.4 The sewerage disposal system is functional and
properly maintained
7.2.1.5 Appropriate ventilation is provided in theatres,
patient accommodation and waiting areas
7.2.1.6 Routine and emergency medical gas and vacuum
systems meet the needs of the health establishment
7.2.2 Operational plant,
machinery and equipment is well
maintained, fully functional and
complies with regulations
7.2.2.1 Operational plant, equipment and installations
are tested and properly maintained
7.2.2.2 The operational plant, machinery and
equipment is upgraded, replaced, decommissioned and
disposed of according to a documented system
7.2.3 A reliable internal and
external telephone system
provides routine and emergency
back-up communication
7.2.3.1 The telephone system is functional and reliable
7.2.4 A functional public
communication system allows
communication throughout the
health establishment in the event
of an emergency
7.3 Safe and
secure
environment
7.3.1 People and property are
actively protected from safety
and security risks
7.4 Hygiene
7.4.1 The buildings and grounds
7.2.3.2 A functional back-up system ensures
communication if the telephone system fails
7.2.3.3 Private telephone facilities are available for
communicating confidential information
7.2.4.1 A system is in place for alerting occupants in the
event of an emergency
7.2.4.2 Staff are briefed to react to emergency warnings
7.2.4.3 All beds and ablution facilities have an
emergency call system to alert the nursing staff
7.3.1.1 Security systems safeguard the building,
patients, visitors and staff
7.3.1.2 The layout of security systems protect
vulnerable patients
7.3.1.3 Adequate internal and external lighting protects
patients, visitors and staff
7.3.1.4 All security incidents are reported and
addressed
7.3.1.5 Safety and security awareness is promoted
among staff
7.3.1.6 Current Local Fire Authority certificates show
the health establishment complies with relevant fire
safety regulations
7.3.1.7 An emergency plan is available to show that
patient well-being is protected at all times
7.4.1.1 The health establishment is kept clean, including
205
Stellenbosch University https://scholar.sun.ac.za
APPENDIX B | NATIONAL CORE STANDARDS
and
cleanliness
are kept clean and hygienic to
maximise safety and comfort
critical areas of public use (especially toilets) and areas
for patient care
7.4.1.2 Appropriate cleaning materials and equipment
are available, and properly used and stored
7.4.1.3 Pests are controlled in internal and external
areas, and infestations are dealt with promptly and
effectively
7.4.1.4 There is a no smoking policy
7.5 Waste
management
7.6 Linen
and laundry
7.7 Food
services
7.5.1 Waste management in the
health establishment and
surrounding environment
complies with legal
requirements, national standards
and good practice
7.5.2 Healthcare risk waste
(HCRW) is handled, stored and
disposed of safely to reduce
potential health risks and to
protect the environment
7.5.3 Management of general
waste (e.g. office, kitchen, garden
or household waste) ensures
general cleanliness and the safety
of staff and patients
7.6.1 Linen and laundry services
meet the needs of the hospital or
clinic and safety standards
7.7.1 Food services are provided
to meet patients’ needs as well as
safety standards
7.5.1.1 There is a current waste management policy and
procedure
7.5.1.2 A designated and knowledgeable staff member
ensures compliance with relevant waste management
legislation and standards
7.5.2.1 The health establishment reviews its HCRW
management every two years to identify the hazardous
waste it generates and establish processes for its safe
management
7.5.2.2 Documented policies and procedures are
available for the collection, handling, segregation,
storage and disposal of HCRW
7.5.2.3 A contract and service level agreement is in
place with an approved and legally compliant waste
removal service provider
7.5.2.4 There are sufficient, accessible and appropriate
waste disposal containers to handle all the HCRW
generated
7.5.2.5 Anatomical waste is disposed of legally while
taking into account cultural preferences
7.5.3.1 General waste is stored and transported
appropriately and securely, and removed promptly
7.5.3.2 Sufficient numbers of suitable containers are
conveniently located to allow safe disposal of waste
7.6.1.1 The laundry service is effectively managed and
delivered (on-site or out-sourced) to meet the needs of
the health establishment and laundry standards
7.6.1.2 All laundry is handled in line with infection
control and safety requirements
7.6.1.3 The laundry has suitable equipment to meet the
needs of the health establishment
7.6.1.4 Adequate stocks of linen are maintained to
ensure that items are always available
7.7.1.1 Policies and procedures guide all aspects of food
procurement, storage, preparation and serving
7.7.1.2 Food services are effectively managed and
delivered (on-site or out-sourced) to meet the needs of
the health establishment
7.7.1.3 Patients are satisfied with food quality and
presentation
7.7.1.4 Food services provide patients with adequate
and nutritious food and drink
7.7.1.5 Policies and procedures are in place for
infection control, safety and food hygiene
7.7.1.6 Food services meet patients’ cultural, religious
and dietary needs
7.7.1.7 Equipment for the safe preparation of food is
available
206
Stellenbosch University https://scholar.sun.ac.za
APPENDIX B | NATIONAL CORE STANDARDS
7.7.1.8 Kitchens meet hygiene and environmental
health standards
207
Stellenbosch University https://scholar.sun.ac.za
APPENDIX C: VERIFICATION OF THE FRAMEWORK
In each of the following tables, the requirements for the layout design framework for rural hospitals
are verified. Verification of this study is done in three ways namely hospital layout design questions,
research objectives, and framework outputs.
C.1 HOSPITAL LAYOUT DESIGN QUESTIONS VERIFICATION
This section checks that all the layout design questions are answered. Table 73 explains how and
where each of these questions is addressed in this study.
TABLE 73: HOSPITAL LAYOUT DESIGN QUESTIONS VERIFICATION
Hospital layout design
questions
What is the capacity of the
hospital?
Which departments should be
included in the hospital layout?
What is the capacity of each
department?
Which rooms are necessary for
each department?
How much space is required for
each room?
Which departments should and
should not be placed in close
proximity to each other?
Where should each department
be located?
How each question is addressed
The capacity of the hospital is suggested according to the
population size of the community and desired occupancy
rate. Additional South African averages are also used.
The compulsory departments for a district hospital were
identified by using the South African laws as guide. There
are also a few optional departments that the user can
choose to include in the layout.
The capacity of each department is suggested by using
benchmarking five hospitals in South Africa. However, the
user is still able to change these values according to
preferences.
The required and optional rooms for each department
were determined using South Africa’s R158.
The minimum area for each room was determined
according
to
South
Africa’s
R158,
provincial
recommendations, other countries’ laws, and industrial
norms. These values were often linked to the amount of
beds or staff where applicable. The user is also able to
adjust these areas and include optional rooms in the
layout.
Muther’s SLP Procedure’s relationship chart was used to
model the relationships between departments according
to the functionality of each department.
The optimal/near optimal location for each department is
determined by optimising an Operations Research method
namely the Quadratic Set Covering Problem with
Simulated Annealing.
Section/s
where
addressed
5.6.1
Appendix A,
4.1.2
Appendix A,
5.6.1
Appendix A,
4.1.2
Appendix A,
4.1.2
3.3, 5.4.7
2.3.4, 2.4.2.2,
5.6.3, 5.6.4
C.2 RESEARCH OBJECTIVES VERIFICATION
This section verifies that all the research objectives (discussed in Section 1.4) of this study are reached.
Table 74 explains how and where each research objective is addressed in this study.
TABLE 74: RESEARCH OBJECTIVES VERIFICATION
208
Stellenbosch University https://scholar.sun.ac.za
APPENDIX C | SOLUTION METHODS FLOW DIAGRAMS
Chapter
Research objectives
How each objective is addressed
Quantitative layout design methods
Identify
the
popular
quantitative layout models
used to design layouts
Analyse
the
various
components of these layout
models
Identify and analyse the
popular methods for solving
these models
Investigate
other
methods such as
software
design
layout
Rural versus urban hospitals
Qualitative layout design methods
and other design considerations
Compare the layout models,
and solution methods with
each other
Identify
the
popular
qualitative design methods
used to plan hospital layouts.
Analyse
the
various
components
of
these
qualitative design methods.
Identify the popular design
considerations for hospitals.
Link the hospital design
considerations
with
the
quantitative methods.
Determine the commonalities
among rural and urban
hospitals.
Determine how laws and
standards affect the design of
a hospital.
The layout models used in articles are
researched. A timeline was also constructed.
Using these results, the more popular layout
models are identified. These layout models were
also classified as discrete or continuous.
For each of the models the main objective(s),
notation, assumptions, inputs, outputs and
variations was analysed.
The solution methods used in articles are
researched. A timeline was also constructed.
Using these results, the more popular solution
methods are identified.
The more popular layout software found in
literature are identified, discussed and classified
as construction or improvement algorithm
based. A timeline is also constructed.
The layout models are discussed and compared
and the most suitable models are proposed. The
solution methods are discussed and compared
and the most suitable methods are proposed.
The more popular approaches used to plan
hospital layouts are identified.
The qualitative design steps are identified and
compared to the discussed layout models. The
most adequate step is selected and applied.
The most important design considerations
applicable to hospital layouts found in literature
are identified. Each one is analysed and layout
implications identified.
The hospital design considerations link up the
quantitative methods analysed in Chapter 2
The commonalities among rural and urban
hospitals are identified by analysing general
building constraints, hospital layout constraints,
and adherence to standards.
The laws and standards impose various
constraints on the hospital layout. These
constraints are identified and incorporated into
the framework.
Section/s
where
addressed
2.2, 2.3
2.3
2.2, 2.4
2.2, 2.5
2.6
3.2
3.2, 3.3
3.4
3.4.6
4.1
Appendix A,
4.1, 4.3
Identify minimum dimensions
and other criteria applicable
to the design of a hospital
layout.
The minimum dimensions and constraints are
identified and incorporated into the framework
using Excel VBA.
Appendix A,
4.1
Determine
constraints.
Rural-specific constraints are identified using
data made available from the South African
government. Rural challenges are discussed via
literature found on the subject.
4.2
rural-specific
209
Stellenbosch University https://scholar.sun.ac.za
Proposed rural hospital design framework
APPENDIX C | SOLUTION METHODS FLOW DIAGRAMS
Determine
the
layout
implications of the ruralspecific constraints.
Describe the real world
problem and how it differs
from the research literature.
Select the most adequate
layout model and solution
method.
Incorporate qualitative design
methods and hospital design
considerations
into
the
framework.
Make necessary adjustments
to
the
framework
to
accurately model the real
world problem.
Development and explanation
of framework.
The layout implications of the rural-specific
constraints are identified and incorporated into
the framework.
4.2, 4.3
The real world problem is described and how it
differs from the research literature is discussed.
5.3
The most adequate layout model is selected
based on a set of criteria namely model
objectives, assumptions, inputs, outputs, and
qualitative design considerations
The selected qualitative design method is
integrated with the selected quantitative design
method. The hospital design considerations are
incorporated into the framework
The flow variable is replaced with the
relationship diagram in order to take the
functionality
and
interdepartmental
relationships of the hospital departments into
account
The framework is developed using Excel VBA,
and RStudio
5.4, 5.5
5.1, 5.4.7
5.4.7
5.6
C.3 FRAMEWORK OUTPUTS VERIFICATION
This section checks that the outputs of the developed framework are realistic and works as expected.
Table 75 shows evidence that the program works in the desired way.
TABLE 75: FRAMEWORK OUTPUTS VERIFICATION
Framework outputs verification
There are no overlapping departments
All the departments are placed within the available space – the
departments centres are within the specified region
The correct number of departments are placed each time
The departments with positive relationships are in general placed closer
together and departments with negative relationships are usually placed
apart
An optimal/near optimal layout solution is reached – the objective
function is optimised/near optimised
210
Evidence
Example 1, Example 2, Example 3,
Example 4, Case study
Example 1, Example 2, Example 3,
Example 4, Case study
Example 1, Example 2, Example 3,
Example 4, Case study
Example 1, Example 3, Example 4,
Case study
Example 1, Example 2, Example 3,
Example 4, Case study
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D: DOCUMENT FOR INTERVIEWS WITH
EXPERTS
This section contains a presentation, documents used for the purpose of validation, and feedback from
experts.
D.1 VALIDATION PRESENTATION
211
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
212
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
213
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
214
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
215
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
216
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
217
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
218
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
219
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
220
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
221
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
222
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
223
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
224
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
225
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
226
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
227
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
228
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
229
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
230
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
231
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
232
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
233
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
234
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
235
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
236
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
237
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
238
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
239
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
240
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
241
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
242
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
243
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
D.2 VALIDATION DOCUMENT AND QUESTIONNAIRE
244
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
245
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
246
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
247
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
248
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
249
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
250
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
251
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
252
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
253
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
254
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
255
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
256
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
257
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
258
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
259
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
260
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
261
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
262
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
263
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
264
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
265
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
266
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
267
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
268
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
269
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
D.3 EXPERT ANALYSIS FEEDBACK
270
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
271
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
272
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
273
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
274
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
275
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
276
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
277
Stellenbosch University https://scholar.sun.ac.za
APPENDIX D | DOCUMENT FOR INTERVIEWS WITH EXPERTS
278
Stellenbosch University https://scholar.sun.ac.za
APPENDIX E: PROGRAM CODE
library("phonTools")
library("functional")
#takes a vector x of positions of departments
#and m is the number of possible positions
#returns the correct matrix version
vectorToMatrix <- function(x , m) {
n <- length(x)
q <- zeros(n,m)
for(i in 1:n) {
q[i,x[i]] = 1
}
return(q)
}
#takes the n by m matrix x and returns
#the vector version
matrixToVector <- function(x) {
n <- nrow(x)
m <- ncol(x)
y <- c(0,0)
for( i in 1:n) {
for (j in 1:m) {
if(x[i,j] == 1) {
y[i] = j
}
}
}
return(y)
}
#returns false if all rotations of departments given by 3D
#array sizes don't fit in the space.
#width and height are width and height in blocks
#y is the vector of positions
BlockMatrixFull <- function(y,width,height,sizes) {
#print(sizes)
possibleRotations <- dim(sizes)[1]
for(i in 1:possibleRotations) {
if(BlockMatrixImproved(y,width,height,sizes[,,i])) {
#print(i)
return(TRUE)
}
}
return(FALSE)
}
#returns false if any departments overlap or don't fit in the space.
#width and height are width and height in blocks
279
Stellenbosch University https://scholar.sun.ac.za
APPENDIX E | PROGRAM CODE
#sizes is a 2D array representing one combination of rotations for the departments
#y is the vector of positions
BlockMatrixImproved <- function(y,width,height,sizes) {
#print(sizes)
sizes_transpose = t(sizes)
covered <- zeros(height,width) #an empty block matrix
count_departments <- length(y)
for (i in 1:count_departments) {
value <- placeDepartment(covered,sizes_transpose[i,],y[i])
if (value != FALSE || length(value) > 1) {
covered = value
} else {
return(FALSE)
}
#print(covered)
}
return(TRUE)
}
#print(BlockMatrixImproved(c(4,3),4,4,array(c(3,2,2,2),dim = c(2,2))))
#print(BlockMatrixImproved(c(1,3),2,4,array(c(2,2,2,2),dim = c(2,2))))
#places a department of width and height given in sizeVector
# at point pointNumber in
# matrixBlock and returns the matrixBlock if it succeeds
# returns FALSE if it fails
placeDepartment <- function(matrixBlock,sizeVector,pointNumber) {
y_center <- floor(sizeVector[1] / 2)
x_center <- floor(sizeVector[2] / 2)
#print(y_center)
widthBlock <- ncol(matrixBlock)
heightBlock <- nrow(matrixBlock)
width <- widthBlock - 1
height <- heightBlock - 1
x_pos <- ((pointNumber - 1) %% width) + 1
y_pos <- floor((pointNumber - 1) / width) + 1
#print(y_pos)
columns <- sizeVector[1]
rows <- sizeVector[2]
for (i in 1:columns) {
for (j in 1:rows) {
y_point <- i + y_pos - y_center
x_point <- j + x_pos - x_center
if(y_point < 1 || x_point < 1 || y_point > heightBlock || x_point > widthBlock) {
return(FALSE)
}
matrixBlock[y_point,x_point] = matrixBlock[y_point,x_point] + 1
280
Stellenbosch University https://scholar.sun.ac.za
APPENDIX E | PROGRAM CODE
if (matrixBlock[y_point,x_point] >= 2) {
return(FALSE)
}
}
}
return (matrixBlock)
}
#print(placeDepartment(matrix(c(0,0,0,0),nrow = 2, ncol = 2),c(2,2),1))#passes
#print(placeDepartment(matrix(c(0,0,0,0,0,0),nrow = 3, ncol = 2),c(2,2),2))#passes
#print(placeDepartment(matrix(c(0,0,0,0,0,0),nrow = 3, ncol = 2),c(3,2),1))#passes
#print(placeDepartment(matrix(c(0,0,0,0,0,0,0,0,0,0,0,0),nrow = 3, ncol = 4),c(2,2),5))#passes
#print(placeDepartment(matrix(c(0,0,0,0,0,0,0,0,0,0,0,0),nrow = 3, ncol = 4),c(3,2),3))
#Works out the QSCP value given number of departments n
#possible positions m as well as
#n*n matrix f
#m*m matrix d
# and the n*m matrix a (or is it m*n?)
FunctionQSCP <- function(y,n,m,f,d,a) {
S <- 0
print("Enter QSCP")
x <- vectorToMatrix(y,m)
#print(x)
for (i in 1:n)
{
for(j in 1:m)
{
S = S + a[i,j] * x[i,j]
}
}
for (i in 1:n){
for(j in 1:m){
for(k in 1:n) {
for(l in 1:m){
S = S + f[i,k]*d[j,l]*x[i,j]*x[k,l]
}
}
}
}
print("Leave QSCP")
return(S)
}
#exchange for any random other number not in the list
#or swap two of the numbers around
ChangeDepartments <- function(x,n,BlockMatrix) {
m <- length(x)
#z <- runif(3)
good<- FALSE
count_watch <- 0
print("Enter Department Change")
281
Stellenbosch University https://scholar.sun.ac.za
APPENDIX E | PROGRAM CODE
while(good == FALSE)
{
count_watch = count_watch + 1
y <- x
z <- runif(3)
if (z[1] >= 0.5) {
#exchange for number not on list
q <- 1:n
q = setdiff(q,x)
#print(q)
selection <- z[2] * (n - m)
selection = floor(selection) + 1
q[selection]
#print(q[selection])
y[floor(z[3] * m) + 1] = q[selection]
#print(y)
good = BlockMatrix(y)
}
else {
index_a = floor(z[2] * m) + 1
index_b = floor(z[3] * (m - 1)) + 1
if(index_b >= index_a) {
index_b = index_b + 1
}
swap = y[index_a]
y[index_a] = y[index_b]
y[index_b] = swap
good = BlockMatrix(y)
#swap two numbers around
}
}
print("Leave Department Change")
#print(count_watch)
return(y)
#return
}
gridPosition <- function(x,y,width) {
#x_pos <- ((pointNumber - 1) %% width) + 1
#y_pos <- floor((pointNumber - 1) / width) + 1
return(((y - 1) * width) + x)
}
main <- function() {
width_blocks <- 20
height_blocks <- 25
positions <- (width_blocks - 1) * (height_blocks - 1)
number_departments <- 16
f <- matrix(c(0,0,0,…, 0),nrow = number_departments,ncol = number_departments)
d <- matrix(c(0,0,0,…, 20),nrow = positions, ncol = positions)
size_array_3D <- array(c(4,4,7,…, 4),dim = c(2,16,65))
#curry the BlockMatrixFull function to give a function that only depends on y
BlockMatrix <- Curry(BlockMatrixFull,width=width_blocks,height=height_blocks,sizes=size_array_3D)
282
Stellenbosch University https://scholar.sun.ac.za
APPENDIX E | PROGRAM CODE
#curry the ChangeDepartments function to give a function that only depends on y
Change <- Curry(ChangeDepartments,n=positions,BlockMatrix=BlockMatrix)
#curry the FunctionQSCPReady function to give a function that only depends on y
FunctionQSCPReady <- Curry(FunctionQSCP,n=number_departments,m=positions,f=f,d=d,a=a)
p = Curry(gridPosition,width=width_blocks - 1)
first_guess<- c(p(18,16),…,p(5,13))
#first_guess is a vector of positions of the departments that HAS to fit.
#To be sure it fits, BlockMatrix(first_guess) HAS to equals true
if(BlockMatrix(first_guess) == TRUE) {
max_iterations = 500
print(optim(first_guess,FunctionQSCPReady,Change,method = c("SANN"),control = list(maxit =
max_iterations)))
} else {
print("error! first guess does not fit!")
}
}
main()
283
Stellenbosch University https://scholar.sun.ac.za
APPENDIX F: UPDATED FRAMEWORK
This section contains the outline of the developed framework after removing the step that pertains to
calculating the optimal layout solution using the Branch and Bound Algorithm. The updated
framework is shown in Figure 33.
1. Inputs
Rural population size
Occupancy rate
Circulation space
Construction cost
Optional departments
5. Hospital departments
5.1 Administration
5.2 Obstetrics
5.3 Operating Suite
5.4 Paediatric Unit
Suggest total number of
beds
Number of hospital beds
Suggest number of beds per
department
5.6 Laundry
5.7 Kitchen
Required
2. Hospital beds
5.5 Outpatient Unit
5.8 Sterilisation & Disinfection Unit
5.9 Pharmacy
5.10 Emergency and Casualty Unit
5.11 Acute Psychiatric Facility
3. Beds per department
5.12 Chronic Care Unit
Number of beds per
department
5.13 Rehabilitation Unit
5.14 Mortuary
Suggest relationships
between departments
5.15 Laboratory
4. Relationship chart
5.17 Neonatal Intensive Care Unit
5.16. Radiology
Optional
Departmental
relationships
5.18 Intensive Care Unit
5.19 High Care Unit
Suggest number of beds
Suggest minimum room areas
Suggest number of single rooms
Yes
EXCEL VBA
RSTUDIO
Number of beds
Number of single rooms
Optional rooms
Size of each room
Calculate total area per
department
Calculate departmental
construction cost
Change inputs
No
7. Simulated Annealing
Calculate optimal layout
using Branch and Bound
method
FIGURE 33: UPDATED FRAMEWORK
284
6. Summary
Area per department
Cost per department
Calculate total hospital area
Calculate total cost